Docstoc

P4P feasibility study final 8-11-07

Document Sample
P4P feasibility study final 8-11-07 Powered By Docstoc
					           Performance-Based Financing


Report on Feasibility and Implementation Options
                               FINAL
                        September 2007




Paul Smithson, Dr Nelly Iteba, Oscar Mukasa, Dr Ali Mzige, Max Mapunda,
                             Gradeline Minja

Assisted by: Mr. Denis Njaila, Mr. Hamisa Mangapi, Mr. Hamisi Matwewe,
 Ms. Hilda Mwabukusi, Nassoro Ally, Ubadius Tumaini, Husna Makemba,
              Freebiki Selemani, Kuluthum Sadiq, Rukia Ally



Study coordinated by Ifakara Health Research and Development Centre, on
behalf of Ministry of Health and Social Welfare, with support from NORAD
Executive Summary
This study examines the feasibility of introducing a performance-related bonus scheme in
the health sector. After describing the Tanzania health context, we define “Performance-
Based Financing”, examine its rationale and review the evidence on its effectiveness. The
following sections systematically assess the potential for applying the scheme in
Tanzania. On the basis of risks and concerns identified, detailed design options and
recommendations are set out. The report concludes with a (preliminary) indication of the
costs of such a scheme and recommends a way forward for implementation.

We prefer the name “Payment for Performance” or “P4P”. This is because what is
envisaged is a bonus payment that is earned by meeting performance targets1. The
dominant financing for health care delivery would remain grant-based as at present.

There is a strong case for introducing P4P. Its main purpose will be to motivate front-line
health workers to improve service delivery performance. In recent years, funding for
council health services has increased dramatically, without a commensurate increase in
health service output. The need to tighten focus on results is widely acknowledged. So
too is the need to hold health providers more accountable for performance at all levels,
form the local to the national.

P4P is expected to encourage CHMTs and health facilities to “manage by results”; to
identify and address local constraints, and to find innovative ways to raise productivity
and reach under-served groups. As well as leveraging more effective use of all resources,
P4P will provide a powerful incentive at all levels to make sure that HMIS information is
complete, accurate and timely. It is expected to enhance accountability between health
facilities and their managers / governing committees as well as between the Council
Health Department and the Local Government Authority. Better performance-monitoring
will enable the national level to track aggregate progress against goals and will assist in
identifying under-performers requiring remedial action.

We recommend a P4P scheme that provides a monetary team bonus, dependent on a
whole facility reaching facility-specific service delivery targets. The bonus would be paid
quarterly and shared equally among health staff. It should target all government health
facilities at the council level, and should also reward the CHMT for “whole council”
performance. All participating facilities/councils are therefore rewarded for improvement
rather than absolute levels of performance. Performance indicators should not number
more than 10, should represent a “balanced score card” of basic health service delivery,
should present no risk of “perverse incentive” and should be readily measurable. The
same set of indicators should be used by all.

1
 “Performance-based financing”, on the other hand, can be used to describe a wide variety of different
systems that relate health care financing to the outputs produced, including service delivery contracts, fee
for payment, fee per case etc. In Tanzania’s case, tax-based financing, provided to councils in the form of
grants, will continue to be the dominant health financing mode. P4P will simply provide the possibility of
earning a bonus if service delivery performance targets are exceeded.


                                                      ii
CHMTs would assist facilities in setting targets and monitoring performance. RHMTs
would play a similar role with respect to CHMTs. The Council Health Administration
would provide a “check and balance” to avoid target manipulation and verify bonus
payments due.

The major constraint on feasibility is the poor state of health information. Our study
confirmed the findings of previous ones, observing substantial omission and error in
reports from facilities to CHMTs. We endorse the conclusion of previous reviewers that
the main problem lies not with HMIS design, but with its functioning. We advocate a
particular focus on empowering and enabling the use of information for management by
facilities and CHMTs. We anticipate that P4P, combined with a major effort in HMIS
capacity building – at the facility and council level – will deliver dramatic improvements
in data quality and completeness. We recommend that the first wave of participating
councils are selected on the basis that they can first demonstrate robust and accurate data.

We anticipate that P4P for facilities will not deliver the desired benefits unless they have
a greater degree of control to solve their own problems. We therefore propose - as a prior
and essential condition – the introduction of petty cash imprests for all health facilities.
We believe that such a measure would bring major benefits even to facilities that have
not yet started P4P. It should also empower Health Facility Committees to play a more
meaningful role in health service governance at the local level.

We recommend to Government that P4P bonuses, as described here, are implemented
across Mainland Tanzania on a phased basis. The main constraint on the pace of roll-out
is the time required to bring information systems up to standard. Councils that are not yet
ready to institute P4P should get an equivalent amount of money – to be used as general
revenue to finance their comprehensive council health plans.

We also recommend that up-to-date reporting on performance against service delivery
indicators is made a mandatory requirement for all councils and is also agreed as a
standard requirement for the Joint Annual Health Sector Review.

P4P can also be applied on the “demand-side” – for example to encourage women to
present in case of obstetric emergencies. There is a strong empirical evidence base from
other countries to demonstrate that such incentives can work. We recommend a separate
policy decision on whether or not to introduce demand-side incentives. In our view, they
are sufficiently promising to be tried out on an experimental basis.

When taken to national scale (all councils, excepting higher level hospitals), the scheme
would require annual budgetary provision of about 6 billion shillings for bonus payments.
This is equivalent to 1% of the national health budget, or about 3% of budgetary
resources for health at the council level. We anticipate that design and implementation
costs would amount to about 5 billion shillings over 5 years – the majority of this being
devoted to HMIS strengthening at the facility level across the whole country.




                                             iii
Acknowledgements
This study was undertaken during the month of July, 2007 and was funded with support
from Norad. We are grateful to the Ministry of Health and Social Welfare for providing
permission to undertake the study and for releasing two members of staff to work with
the consultancy team. Many individuals in central government ministries generously
provided their time, experience and expertise. Although they are too numerous to name,
we express our appreciation to them all. The same sentiment is extended to the many key
informants who facilitated our discussions with health managers and front-line health
workers in Dar es Salaam, Coast and Morogoro regions. The names of all people
interviewed can be found at Annex 1.

As team leader, I would like to extend special thanks to the team for their enthusiasm and
effort. Without this, it would not have been possible to undertake such a complex
assignment in such a short time frame.

Notwithstanding the close co-operation of Norad / Norwegian Embassy and the Ministry
of Health and Social Welfare, the views expressed in this report are those of the authors
and do not represent the policy or official position of the Government of Tanzania or
Norway.

Although we have made every effort to check our facts and eliminate oversight, we are
only human. Any errors of commission or omission remain the responsibility of the
authors.




                                            iv
Acronyms
ACT        Artemisinin Combination Therapy
ANC        Ante-Natal Care
ARV        Anti-Retroviral
CBD        Community-Based Distributor
CCHP       Comprehensive Council Health Plan
CHMT       Council Health Management Team
D-by-D     Decentralisation by Devolution
DED        District Executive Director
EMOC       Emergency Obstetric Care
EPI        Expanded Programme on Immunisation
FANC       Focused Antenatal Care
FP         Family Planning
GFS        Government Financial System
HF         Health Facility
HMIS       Health Management Information System
HSSP II    Health Sector Strategic Plan (2)
ITN        Insecticide-Treated Net
LGCDG      Local Government Capital Development Grant
LGMD       Local Government Monitoring Database
MCH        Maternal and Child Health
MDG        Millennium Development Goal
MMAM       Primary Health Services Development Programme, 2007-2017
MNCH       Maternal, Neonatal and Child Health
MOF        Ministry of Finance
MOHSW      Ministry of Health and Social Welfare
NGO        Non-Governmental Organisation
NHIF       National Health Insurance Fund
NTPI       Norway-Tanzania Partnership Initiative
OPD        Outpatient Department
P4P        Payment for Performance
PMO-RALG   Prime Minister’s Office – Regional Administration and Local Government
PNC        Post-Natal Care
PO-PSM     President’s Office – Public Service Management
SASE       Selected Accelerated Salary Enhancement
TB         Tuberculosis
VHW        Village Health Worker
WDC        Ward Development Committee
WFA        Weight-for-Age




                                      v
Table of Contents

Executive Summary ............................................................................................................ ii
Acknowledgements............................................................................................................ iv
Acronyms............................................................................................................................ v
Table of Contents............................................................................................................... vi
1. Introduction.................................................................................................................- 1 -
   Policy Context.............................................................................................................- 1 -
   NTPI............................................................................................................................- 2 -
2. Pay-for-Performance...................................................................................................- 4 -
   Definition ....................................................................................................................- 4 -
   Examples.....................................................................................................................- 4 -
   Evidence......................................................................................................................- 5 -
3. Applicability in Tanzania............................................................................................- 6 -
   Precedents ...................................................................................................................- 6 -
   Acceptability ...............................................................................................................- 7 -
   Desirability..................................................................................................................- 7 -
   Feasibility..................................................................................................................- 10 -
   Data Quality Audit....................................................................................................- 11 -
   Summary ...................................................................................................................- 12 -
4. Design Options and Recommendations ....................................................................- 14 -
   Level of Services.......................................................................................................- 14 -
   Non-Govt. Providers .................................................................................................- 14 -
   Who Pays Whom?.....................................................................................................- 15 -
   Dealing with Financial Risk......................................................................................- 15 -
   Team or Individual Bonus ........................................................................................- 16 -
   Allocation to Individuals...........................................................................................- 17 -
   Performance-Payment Link ......................................................................................- 17 -
   Selection of Performance Indicators.........................................................................- 18 -
   Who Sets Targets, Verifies Performance? ................................................................- 20 -
   Data quality & verification .......................................................................................- 20 -
   Calculating Performance and Payments ...................................................................- 21 -
5. Measurement in Practice...........................................................................................- 22 -
   Suggested Performance Indicators............................................................................- 22 -
   HMIS Functionality ..................................................................................................- 24 -
   Local Government Monitoring Database..................................................................- 28 -
6. Implementation Arrangements..................................................................................- 30 -
   Supply-Side; Demand-Side, or Both?.......................................................................- 30 -
   Universal, Phased or Experimental...........................................................................- 30 -
   Complementary Measures to Raise Performance .....................................................- 31 -
   Imprest for Health Facilities .....................................................................................- 31 -
7. Indicative Costs.....................................................................................................- 33 -
8. Conclusions and Recommendations .........................................................................- 35 -
   Next Steps .................................................................................................................- 36 -
Annex 1: People Consulted...........................................................................................- 37 -
Annex 2: References .....................................................................................................- 39 -


                                                                  vi
Annex 3: Functionality of HMIS ..................................................................................- 44 -
  Introduction...............................................................................................................- 44 -
  Reviews and Remedial Action..................................................................................- 44 -
  Organisation..............................................................................................................- 45 -
  Inputs and logistics ...................................................................................................- 46 -
  MTUHA TOOLS ......................................................................................................- 47 -
  Information Flow ......................................................................................................- 48 -
  HMIS in Practice.......................................................................................................- 48 -
Annex 4: Experience and Evidence on P4P..................................................................- 50 -
Annex 5: Terms of Reference .......................................................................................- 56 -




                                                               vii
1. Introduction
This report examines the potential for the introduction of “performance-based financing”
(or “pay-for-performance”) in the Tanzania context as a key element of Norway’s support
to Tanzania under the NTPI.

Performance-based financing is gaining growing attention as a means of achieving
greater results-focus, encouraging local initiative to solve problems, finding new ways of
reaching under-served populations, raising productivity and improving quality. The
proposition is that financial rewards can be used to encourage and reward such
behaviours. Various forms of pay for performance have been implemented in a wide
variety of contexts, including developing countries with relatively weak health systems.
Although formal evaluations are scarce, early experience has been encouraging. The
concept has already been taken to national scale in Rwanda.

This report begins by setting out the policy context in Tanzania, the aims of the NTPI,
and the rationale for “Pay for Performance” (P4P). Chapter 2 looks briefly at experience
and evidence on this concept in other settings. Chapter 3 examines whether such a
scheme is applicable in the Tanzania context. With reference to the various concerns
raised by key informants, Chapter 4 works through the detailed design considerations and
provides options and recommendations on how it should be operationalised. Chapter 5
looks in greater depth at potential indicators, the current capacity to measure these, and
proposals to assure data quality. Chapter 6 sets out broad implementation options,
including monitoring and evaluation. Chapter 7 provides a preliminary indication of the
cost of implementation in Tanzania. The final chapter summarises conclusions and
recommendations for the way forward.

Policy Context

Tanzania is committed to meeting the Millennium Development Goals. This includes
Goal 42 (reduce child mortality) and Goal 53 (improve maternal health). Tanzania’s long-
term development goals are also enshrined in Tanzania Development Vision 2025. Over
the medium term (-2010), more specific goals are set out in the National Strategy for
Growth and Reduction of Poverty or “Mkukuta”. Health-sector specific goals are
described in the Health Sector Strategic Plan, which is due to be reviewed and revised for
the period from 2008/09, drawing upon the lessons learned from Health Sector
Evaluation. A new National Health Policy has just been approved by Cabinet, as has a
new strategy for strengthening primary health care4. With regard to MDGs 4 and 5,
Tanzania has drawn up a “Roadmap for Maternal, Newborn and Child Health”.



2
  Reduce by two-thirds the mortality rate of children under five by 2015
3
  Reduce by three-quarters the maternal mortality ratio by 2015
4
  Mpango wa Maendeleo ya Afya ya Msingi (MMAM) or Primary Health Services Development
Programme, 2007-2017


                                             -1-
Table: Specific Goals Relating to MDGs 4 and 55
Goal / Target            Baseline         Latest Estimate      Mkukuta Target         MDG Target
                                                                   2010                 2015
Maternal                   529                  578                265                   132
Mortality Ratio           (1996)              (2004/5)
Under-Five                 145                  112                   79                   48
Mortality                 (1991)              (2004/5)

As the table illustrates, dramatic progress has recently been made with regard to child
mortality rates. In fact, annual disaggregation of the latest data reveals that under-five
mortality in 2004 had already reached 83 – close to the Mkukuta target levels. The MDG
target is within reach if the pace of recent progress can be sustained. While the recent
data on under-five mortality is encouraging, it is striking that most of the improvement
has occurred in “post-neonatal mortality”. The latest data suggest modest (but not
statistically significant) improvement in Neonatal Mortality. This now accounts for
nearly half of all infant deaths, and almost a third of under-five deaths.

Neonatal mortality is intrinsically linked with maternal health. It is therefore not
surprising to find that Maternal Mortality also shows no improvement since the previous
estimate 10 years earlier – or may even be worse. There has also been little improvement
in the proportion of births attended by skilled health personnel or the proportion of
deliveries being carried out in health facilities and the situation in rural areas is markedly
worse. Recent surveys6 have highlighted the fact that the capacity to provide basic
services to save mother’s lives (Emergency Obstetric Care) is largely absent in health
facilities below the hospital level. Intensive effort will be needed to make improvements
in maternal health – and this in turn holds the key to making progress on averting
neonatal deaths and sustaining progress on MDG4.

NTPI
Norway and other partners at the international level have been developing a global
business plan to accelerate progress on MDGs 4 and 5. This will translate into specific
support for a number of countries, including the “Norway Tanzania Partnership
Initiative” (NTPI) in Tanzania. Talks between Norway and Tanzania began between the
Minister of Health and the Norwegian Prime Minister’s Office in December 2006. Talks
culminated in the signing of a Joint Statement between the Governments of Tanzania and
Norway, signed by the respective Heads of State during President Kikwete’s visit to
Norway in February 2007. Subsequently, extensive dialogue with government and other
development partners – including a workshop in April - has formed the basis for a draft
Programme Document that sets out in more detail how Norway’s assistance will be
manifest.

Key design considerations included:
      Accelerating progress towards MDG4 and 5 in Tanzania
      Working with and through government systems and structures
5
 Latest estimates are official figures from the Tanzania Demographic and Health Survey 2004-05
6
 Situation Analysis of EMOC for Safe Motherhood in Public Health Facilities in Tanzania Mainland
(2006); Tanzania Service Provision Assessment Survey 2006 – Preliminary Report (2006).


                                                -2-
       Using joint financing mechanisms
       Channelling resources towards front-line essential services
       Increase emphasis on accountability for enhanced performance
       Explore the potential application of “performance-based financing” in the
       Tanzania context

Specifically, the programme is expected to include:
       Additional funding to improve districts health services channeled through the
       pooled basket fund.
       Facilitate introduction and scaling up of result/output-based financing schemes for
       district level health services including MNCH interventions.
       Improve the quality of and use of the health information systems.
       Contribute to scaling up of community based strategies to promote healthy
       behavior during pregnancy, childbirth and in the postpartum period.

Approximately 80% of the total assistance is expected to be channelled through the
district health basket, amounting to an additional amount of approximately $5m
equivalent per year. This would raise the mean basket fund for councils from around
$0.75 to $0.90 per capita per year. The remaining 20% is expected to cover
complementary components including Monitoring and Evaluation, specific support to
strengthen the Health Management Information System, and support for innovative
approaches through Non-Governmental providers.




                                          -3-
2. Pay-for-Performance
Definition
The term “Pay-for-Performance” of “P4P” is used to describe “Transfer of money or
material goods conditional on taking a measurable action or achieving a predetermined
performance target.”7 This very broad definition can actually cover a multitude of
mechanisms with quite different characteristics.

There are two broad categories of P4P systems. One is a supply-side P4P that provides
incentives to health providers to increase performance. The second is a demand-side
incentive that encourages desired behaviours on the part of the health consumer.
Although the terminology might seem new, the basic rationale is quite simple and
examples already exist in Tanzania.

Examples
On the supply side, examples include:
   • Moving from grant-based funding to variable funding, depending on outputs. This
       is the intention of the “service agreement” arrangement with non-government
       health providers, where hospitals would be paid according to their output, instead
       of according to a “bed grant”
   • Paying providers according to the cases they have treated, with a standard
       reimbursement for clinical categories. This is the system operated by the NHIF to
       reimburse health providers for services rendered to NHIF members.
   • Providing a material reward (computers, printers, fax) to the best-performing
       councils according to immunisation coverage, as has been done by the EPI
       programme.

Some sort of linkage between the amount of funding provided and the amount of services
delivered is actually the rule rather than the exception. All insurance-based health care
systems operate in this way, as do some tax-financed systems (eg the purchaser-provider
split in UK). In the private sector, “merit pay” or “performance-related pay” has also
been a feature for a very long time. Outside of developing countries, it is quite unusual
for a health provider to be paid the same amount, regardless of the amount of output
produced.

On the demand side, examples include:
   • Providing cooking oil to mothers who brought their children to MCH clinics (as
       was done in the 1970s)
   • Providing an ITN voucher to women who attend ANC services

Other examples on the demand-side include providing material incentive to TB patients
to complete their course of therapy, or nutritional support to HIV patients to encourage
treatment compliance (and make it easier to take the medicines).
7
    Eichler, R (2006)


                                           -4-
There is a growing interest in exploring the potential for P4P in health care in developing
countries. At the international level, this is represented by “results-based aid”, of which
GAVI’s funding for immunisation is one example. Countries are also exploring the
possibility of using material incentives to inspire and enable health providers to achieve
higher levels of productivity.

Evidence
One country in the region (Rwanda) has already taken this to national scale, having
started with pilot schemes in 3 provinces about 5 years ago. Rwanda’s “approche
contractuelle” is credited with a dramatic increase in health system productivity. First-
hand observation8 of the scheme is even more convincing than the data. Health workers
at the health facility level described how they had used their own initiative (and the extra
money) to:

    •   Alter opening hours to make it more convenient for clients to attend
    •   Reduce or waive fees
    •   Reward TBAs with token cash payments for referring women to deliver at the
        health facility
    •   Discover – through consultations with their clientele – what needed to be done to
        services to make them more attractive

“P4P” has been tried in many different settings, including post-conflict countries
(Cambodia, DRC) – showing that it can be applied in less advanced health systems. In
spite of the positive results that have been reported in different country examples, there is
very little formal, empirical evidence. This is particularly true of “supply-side”
incentives. Evidence of the effectiveness of demand-side incentives is much stronger –
with a number of controlled experiments demonstrating improvements – for example – in
TB treatment completion. An annotated bibliography describing some of the studies and
results from different country settings can be found at Annex 4.

To the extent that some common lessons emerge from these studies, they show that:
    • The incentive must be applied to actors who are in a position to make a difference
    • Incentives must be designed carefully to avoid undesirable results
    • Incentives must be perceived to be transparent and “fair”, otherwise friction,
       jealousy or a sense of injustice will outweigh the benefits
    • Performance measurement must be seen to be objective, transparent and timely
    • A properly designed supply-side incentive does have the effect of focusing
       providers on achieving results
    • Where constraints lie on the demand-side, provider incentives may make little
       difference. What is needed here is demand-side incentives



8
 The lead author of this study, the Director of Policy and Planning (MOHSW) and the Head of Health at
CSSC made a short study tour to Rwanda earlier this year.


                                                 -5-
3. Applicability in Tanzania
Precedents
The idea of providing material reward for excellent performance in the Tanzania health
sector is not a new one. The EPI programme as a whole receives performance-linked
financing from GAVI, and was rewarded with an additional $3m for improvements in
immunisation coverage since 2002. Sub-nationally, the best performing councils were
rewarded in-kind with computers, printers and fax machines, while additional resources
were also devoted to bringing up the poorest performers. Similarly, the idea of trophies or
certificates of commendation has been used in the past. The national “Guidelines for
Reforming Hospitals at Regional and District Levels” also include the concept of rewards
for the best performing facilities.

Conditionalities linked to the Local Government Capital Development Grant could also
be construed as “performance-related funding” since councils need to meet certain
minimum criteria (mainly linked to financial management) to qualify for the grant. In the
case of the Health Basket Fund, councils must have approved Comprehensive Council
Health Plans and must also submit quarterly financial and technical reports to trigger the
biannual release of funds. This is widely regarded as having had a dramatic impact in
improving compliance to financial management standards, with the number of qualifying
councils growing rapidly since its introduction.

At the individual level, the Selected Accelerated Salary Enhancement scheme (SASE)
provided a salary top-up for civil servants in several sectors. The top-up was explicitly
linked to individual performance agreements which set out an individual’s work plan and
objectives. In practice, however, the SASE experience has not been a happy one. The top-
up was routinely paid, regardless of individual performance – undermining the credibility
of the pay-performance linkage. Moreover, the fact that only some staff benefited from
the scheme created division and dispute. Both shortcomings have important lessons for
the design of future initiatives. The scheme is due to come to an end this financial year.

There are also examples of incentives on the demand-side. In the 1970s, nutritional
supplements (cooking oil etc.) were provided to mothers who brought their children for
under-five clinics. More recently, the provision of ITN vouchers under the Hati Punguzo
scheme provides an incentive for pregnant women to attend ANC and for infants to
attend measles vaccination. In both cases, these initiatives combine an attractive incentive
with something that conveys direct health benefits. The provision of food packages to
patients on ARV works in a similar way – encouraging treatment compliance while also
providing vital nutritional support.

Recent years have witnessed growing interest in Tanzania and elsewhere for welfare
payments to the most destitute. Such “social protection” schemes are designed primarily
to support income for the most vulnerable. However, they may also be linked to school
attendance and preventive health measures as a “conditional cash transfer”, as in Mexico.



                                           -6-
Acceptability
The disappointing experience with SASE has made government nervous about individual
cash incentives for performance. Respondents pointed out that the various failings of
SASE (ability to maintain credible linkage with performance; inclusion and “fairness”)
would need to be addressed if such as scheme is to be instituted. We understand that
“group” or “team” incentives are seen in a more favourable light, and that incentives in-
kind are less controversial from a pay-policy standpoint than cash payments.
Nonetheless, there is some concern that it would still be seen as unfair if a scheme like
this was operated for one sector (health) and not others. If such a scheme is to be
instituted, it is clear that it would require the active support President’s Office - Public
Service Management.

Desirability
As we have seen from the precedents already described, the notion of providing
recognition and material reward for excellence is already well-accepted in the health
sector. Central MOHSW staff generally favoured in-kind rewards rather than cash
rewards, particularly if these are “reinvested” in service delivery. The difficulty with this
logic is that “winners” get extra support for service delivery, while “losers” don’t.
Conversely, if “losers” obtain special remedial support, the scheme might perversely
encourage poor performance.9

The literature is clear that the incentive needs to convey direct benefit to individuals
(rather than the health insitution) if it is to have any motivational effect. This was
confirmed in interviews with CHMTs and front-line health workers. They were
unanimous that the incentive must ultimately go to health workers to have the desired
impact, and that cash benefits are preferable to “in-kind”.

At the same time, all respondents (central government and local government) were of the
view that health service delivery requires a team effort, and should be rewarded as such.
All health staff – including the cleaner – contribute in one way or another to improving
service delivery. Accordingly, it was recommended that performance should be measured
and rewarded for the whole facility, while the incentive would then be distributed to the
staff working there10. This is a similar model to the one operational in Rwanda. Such a
team-based reward should help to avoid disputes over “unfairness” and “inclusiveness”
as long as the rules for the distribution to individual staff are also viewed as clear and
fair.

A number of other concerns were raised in interviews. Foremost among these was the
notion that direct competition between facilities and councils for incentives would be

9
  This “perverse” incentive was recognized in EPI, where under-performing districts received extra
training. Some programme staff are concerned that the associated per diems may actually encourage
councils to under-perform(!).
10
   Respondents in the councils also recommended including community-based outreach workers (VHWs
and CBDs) since their efforts are vital to reaching people in the community and encouraging uptake of
essential services. This is all the more important considering that these staff currently receive no
remuneration.


                                                 -7-
unfair because some places would find it much easier than others to attain the targets. The
main reason for this is wide variation in availability of skilled staff. There are also
variations in condition of health infrastructure, equipment and supplies, as well as
contextual factors (communications, roads, population density, levels of education etc).
These considerations point towards designing an incentive system that rewards
improvement rather than setting a single “hurdle rate” that every facility must attain.

Related to this is the assumption by respondents that P4P would mean that some councils
would get paid more than others – and that this would be in conflict with the formula
basis for local government grants that has been instituted in recent years. We do not think
that this should necessarily be the case if the scheme is implemented on a decentralised
basis. Under this scenario, councils operate the scheme by setting aside x% of their
basket fund grant for the purpose of performance incentives. In case the targets are not
met, any unspent amount would be retained and “rolled over” into the following planning
year, providing additional funds for addressing service delivery constraints and making it
more likely that the targets were attainable in the following year.

Similarly, there was common feeling that supply-side constraints (staff, skills,
infrastructure, equipment, supplies) need to be addressed before a performance bonus
scheme could work. Yet this is precisely one of the objectives of P4P. By focusing
collective attention on results, front line health workers and their managers are forced to
consider what needs to be done in order to achieve the targets - it is not simply a case of
“working harder”. Moreover, the amount of money to be allocated to P4P will be a very
small proportion of the overall budget (roughly 1% of total public expenditure on health,
or about 3% of health spending at the local government level). Thus very sizeable
resources (the other 97%-99%) are available for solving supply-side problems. The
expectation is that P4P will catalyse most effective use of other budgetary resources by
encouraging staff and managers to identify and address the most binding constraints in
their own specific circumstances.

Another concern voiced by most respondents was sustainability. If this scheme is funded
with Norwegian money, will it simply cease at the end of the programme? And if it
ceases, might we not be worse off than before by raising and then dashing expectations?
This is a valid concern. However, if the scheme is funded from basket resources (rather
than earmarked funds from Norway), it could be sustained using that resource. If
experience shows that P4P provides a cost-effective means of raising productivity, there
is no reason why it could not be funded out of budgetary resources in the future, as long
as the scheme is not prohibitively expensive.

Related to the concern above, there is a theoretical risk that P4P could undermine health
workers sense of duty and work ethic. We recognise that the intrinsic motivation and
work ethic of health workers is hugely valuable and is the primary motivator11. Might
P4P supplant this, fostering a culture that staff would do even less work than before if the
scheme ceased? On balance, we do not consider this to be a major risk. We do think that

11
 We came across numerous examples of health workers reporting that they work overtime, nights,
weekends and even use their own money to deliver services.


                                                -8-
performance might drop back down if the scheme ceased, but we do not think that an
incentive at the margin would actually undermine a widely-held sense of duty to patients
– rather it should reinforce this by encouraging providers to attract patients by being more
responsive to client perceptions.

Conversely, there is a risk that the “incentive” might come to be seen as just another
aspect of the workers’ entitlement. This was the issue with SASE and is a common
feature of individual “merit-based pay” schemes in public and private sectors. It often
occurs where managers are asked to make a subjective judgement on the performance of
their staff and who authorise the bonus to avoid conflict. If performance bonuses are
“routinely” paid, it is quite clear that any credible linkage with performance will indeed
be lost, and the net result is simply an inflation of the wage bill. These considerations
highlight the importance of making the performance requirements sufficiently
challenging12, making the measurement of performance as objective as possible, having a
non-discretionary linkage between performance and pay, and communicating clearly the
purpose, rationale and rules of the scheme.

There is also a risk (generally not raised by people interviewed) that P4P could actually
have “perverse” (unintended and damaging) consequences. For example, it could be that
workers will focus only on those services that are being measured and rewarded,
neglecting other essential elements of their duties. This highlights the importance of
selecting measures of service performance that encourage delivery of a comprehensive
package of essential services. Another example might be rewards that potentially
encourage inefficient or unnecessary clinical intervention – the most infamous case being
the link between provider-payment and unusually high caesarean section rates in Brazil.
In Mexico nutrition assistance for underweight children actually worsened malnutrition –
possibly because only malnourished children qualified and mothers wanted to sustain
their entitlement.

Another consideration (again, not generally raised by respondents) is the degree of
freedom that health facilities have to make a real difference to performance. Certain
actions and innovations (like providing 24-hour cover; opening at weekends) are clearly
feasible at the local level. However, what is a facility team to do about things beyond
their immediate control (number of staff, availability of equipment, authority and
resources to undertake maintenance, availability of petty cash to pay for service
enhancements)? If facilities are largely lacking any freedom, authority or resources to
solve local problems, how can we expect incentives to make any difference? It is striking
that in Rwanda (where incentives are associated with imaginative innovations and
dramatic performance improvement), health centres hold a bank account, retain user
fee/insurance revenues, and are able to hire contract staff. In Uganda, a trial compared
facilities that were offered performance bonuses with those that were simply given
greater freedom over how they used government grants. Even the latter achieved
significant improvements, pointing towards the importance of combining the will to


12
 In other words, performance targets should be sufficiently ambitious that a typical health facility cannot
meet all of its targets, all of the time.


                                                   -9-
improve with the freedom to do so. Without greater control over resources, it seems
unlikely that performance bonuses at the facility level would work.

Finally, while P4P could be a powerful tool in enhancing performance, it could also
subvert accountabilities. If, for example, performance bonuses were paid to facilities
directly by MOHSW, it would undermine the accountability of the facility to the CHMT
and to local government. The systems will need to be designed in such a way that they
reinforce – rather than undermine – accountability to local government. Conversely, the
system could be designed in such a way as to reinforce the accountability of health
service providers to local government, including village governments, facility
committees, the WDC, the council health management team, the council administration,
and Council Assemblies.

Feasibility
The feasibility of a performance-based bonus system rests on the ability to report credible
and timely performance data. This is probably the greatest single challenge to the
implementation of P4P in Tanzania. Issues relating to indicators, measurement and
reporting are discussed in greater detail in Chapter 5.

Experience in other countries has shown that measurement of performance must rely
upon routine data systems. Survey-based information collection is expensive, cannot be
carried out with sufficient frequency, and is anyway subject to confidence intervals that
cast doubt over whether or not targets have been met. This means that whatever
performance measures are selected must be measurable from the routine data system.

Reporting of routine data is currently subject to delays, errors and omissions. In the
course of our fieldwork, we have attempted to make quantitative estimates of the
completeness of data reporting from facility – CHMT – Region – National level
(described further in Chapter 5).

Our findings indicate that the problem does not lie primarily in the recording of primary
data in the daily registers and tally sheets. For the most part, staff are remarkably diligent
in filling these out. 13 The main problem seems to begin with the summarising of these
source data into monthly and quarterly summaries for onward transmission to the CHMT.
With rare exceptions (e.g. EPI) data is not interpreted and used at the facility level,
reducing the incentive to analyse it and assure data quality. No national-scale training for
front-line health workers on HMIS and data interpretation has been conducted since
1994-7. CHMT supervisors do inspect registers and try to follow up missing reports.
However, they do not typically provide any feedback on the information obtained, nor do
they use it for performance assessment across facilities. The new data management tool
for CHMTs allows the entry and generation of standard reports (at whole council level).
But it does not preserve the facility identity of the information so obtained, nor does it

13
  The general impression is that the recording of primary data in registers is worse for some registers than
others eg Laboratory, Inpatient and Surgery registers are less likely to be complete than the routine MCH
and OPD registers. There is also a clear consensus that OPD registers are less likely to be filled in
accurately and completely in busy facilities with multiple consulting rooms than in smaller facilities.


                                                   - 10 -
allow interpolation for missing data. Onward transmission of data files to Regional level
are therefore subject to an unknown level of data omission. The same applies at National
level (although it is known how many regions have reported). Modifying the existing
database to preserve the facility identity of data entered and monitor the completeness of
the reports at each level would be a relatively simple task. Similarly, it would be a
relatively simple matter to generate standard performance reports for feedback to
individual facilities so that they can compare themselves to national standards and to
other facilities.

In brief, the main challenges are not so much system problems as human ones.
Stakeholders at every level need to understand and be encouraged to use data for decision
making. A new culture needs to be inculcated so that performance-reporting is regarded
as just as essential as financial accountability.

We do believe that P4P can be introduced in a limited number of councils where
reporting compliance is already high and where routine data is reasonably complete and
credible. We also believe that the P4P system will be a powerful incentive for improving
reporting compliance and accuracy. In the short run, the choice of performance indicators
must be limited to those where data is routinely transmitted from facility to CHMT level
through Book 10 (Form 004) and through parallel systems (eg EPI). In the medium term,
if stakeholders wish other indicators to be routinely measurable, then standard quarterly
report content must be modified to accommodate them, as long as the source data can be
harvested with little difficulty from registers and tally forms.

Whatever the case, it is very clear that the introduction of P4P will require a significant
investment in the HMIS – not to reinvent it from scratch – but to make it useful at facility
and CHMT levels, and to significantly raise reporting completeness, accuracy and
timeliness.

Data Quality Audit
P4P introduces a risk that data could be deliberately inflated in order to qualify for
performance bonuses. The only way that this risk can be mitigated is through some form
of data audit. This would need to stand a good chance of picking up any significant
manipulation of data. In the case of EPI, the introduction of the GAVI performance-based
funding included periodic data quality audit. This entails a sample-based check on data
submitted, checking summary data against source books, and verifying a smaller number
of source book data against children’s immunisation cards. While this does provide a
robust, independent check on data quality, it is also expensive. The solution adopted by
the Tanzania EPI programme was to introduce a routine “internal audit” of the data, using
staff at the council and regional level. A system such as this would help to weed out
obvious errors, reduce the size of the external audit sample required, and ensure that
facilities know that there is a high probability that data falsification will be detected.

As regards data errors, it should be possible to put in place a system that identifies
suspect data, for example through the use of “range limiters” in the input field of the
district HMIS database, and through the identification of anomalies and “outliers”.


                                           - 11 -
Summary
With very few exceptions, respondents were
very enthusiastic about the idea of                  At Mwendapole dispensary in Kibaha
                                                     District staff were asked what actions
performance-related incentives. They did feel        they could take in order to increase
that such incentives could and would inspire         service uptake and win their
greater productivity and greater focus on            performance bonus. They immediately
actual results. They also recognised that such       suggested three simple initiatives that
a scheme could leverage more effective use of        they would take:
                                                     • Open on weekends
all budgetary resources (by identifying and
                                                     • Maintain a 24-hr duty roster
solving the most binding constraints) and            • Use Village Health Workers and
could empower health workers to innovate                 Community leaders to mobilize
and find local solutions to local problems.              clients to use services
Various concerns and risks were raised by
people interviewed. This list of risks is further expanded on the basis of the literature and
experience in other countries. The key risks are summarised below. Many, if not all of
these risks can be addressed through careful design and the design choices recommended
in the next chapter directly respond to the risks and concerns identified.

Summary of Potential Risks
 Risk                      Mitigation
 1. Not allowable          Engage PO-PSM, MOF, PMO-RALG from earliest stage of design
 2. Level playing field;   Use P4P to reward improvement compared to baseline, not
 winners & losers          absolute level of service delivery
 3. Unfair (between        Reward whole team for health facility performance. Clear rules for
 individuals)              sharing bonus across staff (equal)
 4. Subvert resource       Provide additional resources to all councils according for formula.
 allocation formula        Some councils introduce P4P, others use extra funds for CCHP
 5. Need to fix the        99% of the health budget still available to fix supply side! P4P will
 health system first       focus remaining resources on fixing the most pressing constraints
 6. No freedom to          Councils already have high level of autonomy. Relax allocation
 innovate                  ceilings for health block grant and focus on outputs instead of
                           inputs. Introduce petty cash imprest for health facilities
 7. Sustainability         Funded from basket, not earmarked Norwegian funds. Could be
                           integrated into government budget later if proven successful
 9. Undermine work         Not if the bonus depends on meeting targets rather than “fee per
 ethic                     unit of service”
 10. Perverse incentive    Careful design of indicators to avoid rewarding perverse behaviour
 11. Undermine local       Use as a local government tool to reinforce accountability to HF
 accountability            committees, village govt., council admin, full council.
 12. Can’t measure         Use only indicators already reported on (or redesign Form 004);
 performance               strengthen HMIS; institute internal and external data quality audit


The major feasibility constraint lies with HMIS. In the short run, P4P can only be
introduced in those councils which already have a demonstrable track record of data
quality, completeness and timeliness. Over time, it should be possible (especially with the
attraction of the P4P scheme) to bring other councils up to standard.

It should be possible to institute a data quality audit procedure that routinely verifies
(“internal audit”) the data submitted by facilities to CHMTs and CHMTs upwards. An


                                              - 12 -
external data quality audit system could then be put in place to provide an independent
check on the effectiveness of the internal controls on data quality. Deliberate falsification
of data should be penalised by withholding the performance bonus.




                                            - 13 -
4. Design Options and Recommendations
In this section we examine in turn the main design features of a P4P system in Tanzania.
For each of the parameters we make recommendations based on experience in other
countries and the specific Tanzania context. The parameters are:

    Which level of services: primary level, secondary, tertiary, specialist?
    Government providers only, or non-government too?
    Who pays P4P to whom?
    How to avoid “winners and losers”?
    Team bonus, individual incentive, or both?
    How to allocate fairly to individuals?
    How to link bonus to performance?
    Which indicators of performance?
    Who sets targets & verifies performance?
    Data quality assurance
    How to make the bonus payments?

Level of Services
We recommend that the scheme be limited to basic health services up to the level of the
district hospital. District hospitals should be included because it would be seen as unfair
to include some council health services and not others, because they provide a substantial
portion of outpatient as well as referral services, and because they are the major provider
of delivery services. Higher level hospitals should be excluded (at least in the first
instance) because their inclusion would stretch “bonus resources” too thinly across a
much larger number of staff, because they would require a different set of performance
indicators, because they are managed by separate authorities from the council health
services, and because negotiation of the performance indicators and bonus package would
likely be protracted. If the pilot with council health services is successful, the scheme
could be introduced to selected higher level hospitals at a later date. In the few councils
where a regional hospital serves as the district hospital, we recommend it should be
included in the bonus scheme, using council resources.

Non-Govt. Providers
In principle, a P4P system could be extended to private-non-profit and private-for-profit
providers. We do not recommend inclusion of the latter because any reimbursement
system would need to reimburse the whole cost of a treatment episode rather than only a
marginal performance bonus. In any case, some of these providers are already included in
a payment system that rewards outputs (the NHIF). We also do not recommend inclusion
of all private-non-profit providers for several reasons. First, a performance-related
payment system (service agreement) has already been developed and negotiated with the
faith-based providers. Adding a performance bonus would require “unravelling” and
renegotiating the service agreement – which has already taken years to design. Because
they are controlled by multiple different denominations and diocese, such a negotiation
would take a long time and would delay the start of the scheme. However, we do


                                           - 14 -
recommend the inclusion of “district designated hospitals” because they are funded (to all
intents and purposes) in an identical fashion to government hospitals and because they
form a recognised and integral part of council health services.

Who Pays Whom?
A P4P system could be envisaged that simply rewards “whole council” performance
without extending to the facility level. We do not recommend this because the CHMT,
without the active participation of front-line health workers, could not transform service
delivery at the facility level.

Nor do we recommend a system whereby central government (whether MOHSW or
PMO-RALG) pays a variable bonus to CHMTs depending on performance. This would
introduce an element of financial risk that would undermine the resource allocation
formula and would mean that some councils got more than others. Secondly, it would
make Council Health Departments accountable for performance to Central Government,
whereas under D-by-D they should be accountable to the Council Administration and the
elected Council Assembly.

What we do recommend is a P4P system that operates entirely at the local government
level. Councils participating in the first wave of P4P would agree to set aside a fixed
percentage of their basket fund for the payment of performance bonuses. All government
facilities (including DDH) would be eligible for performance bonuses according to their
individual performance. In addition, the council health department administration would
be eligible for bonus depending on the performance of the council as a whole. This would
motivate both front line health workers and the council health department on whom the
facilities depend for supplies, equipment, rehabilitation works and staff.

Dealing with Financial Risk
A performance bonus by definition entails a degree of uncertainty. The expenditure will
only be made if the facilities and councils meet their performance targets. As discussed
above, if the performance funds were paid out from Central Government, the net result
would be to undermine the resource allocation formula since some councils would get
more than others.

We therefore recommend that all councils would get the additional basket resources
resulting from Norway’s contribution to the basket fund – raising mean “district basket”
allocation from $0.75 to $0.90 per capita. Those councils not participating in the first
wave of P4P would still get the extra resources (thus preserving the integrity of the
resource allocation formula). In their case, the funds would represent an expansion of the
resource envelope available for service delivery – but without a hard performance-
linkage. These non-participating councils could join the P4P at a later date, depending on
the lessons learned from the first wave and their level of readiness.

Those councils participating in the first wave of P4P would set aside a fixed percentage
of their basket funds. This would ensure that the amount allocated for bonuses was
proportional to the population being served (plus the other weightings of poverty, burden


                                           - 15 -
of disease and route distance). It would also prevent “inflation” of the funding allocation
to performance bonuses.

The amount of performance bonus to be paid out quarter by quarter would depend upon
actual performance. In case facilities do not meet their targets and a portion of the money
remains unspent, we recommend that the funds remaining be retained in Account No. 6
and rolled over to the following year to be used as general revenues for funding the
Comprehensive Council Health Plan. The funds should not be rolled into the allocation
for performance bonuses in the following year. In other words, once facilities fail to meet
their targets in one year, they cannot “win” the money back by meeting their targets in
the following year. Each year’s performance bonus fund remains a fixed proportion of
basket fund.

One implication of the “fixed proportion” rule would be that the level of the individual
bonus would be higher for councils with fewer staff. We view this as a positive design
feature, helping to provider extra incentive and reward to staff who work in the least
popular councils.

Team or Individual Bonus
Experience with individual bonuses in public and private sectors has been fraught with
difficulty. A bonus based upon individual performance (usually called “merit pay”, or
“performance-based pay”) often creates jealousies and friction in the work place. Added
to this is the subjectivity of performance measurement through appraisal which reduces
transparency, runs the risk of “grade inflation” and runs the risk of the bonus becoming
regarded as “entitlement”.

A further problem is that many aspects of individual performance depend on the
contribution of others (team work). In the health sector, team work is absolutely vital and
it is the performance of whole facility teams that we should be rewarding and
encouraging. We therefore recommend that the qualification for bonus payment is based
solely and wholly on the output of the facility as a whole, as objectively measured by
service delivery statistics. The resulting “team bonus” should then be paid out to
individual staff according to a clear and unequivocal guideline. We recommend that
village health workers or other outreach workers be included as eligible for payments
because they can make a major contribution through “outreach” services, community
health education and mobilising patients to attend the clinic.

At the CHMT level, we recommend that a bonus is paid according to the performance of
the whole council against the “whole council” service delivery/coverage targets. This will
encourage CHMTs to set individual facility targets that are consistent with their overall
council targets (and vice versa). It will also encourage them to ensure that any constraints
at the facility level that require action by the Council Health Administration are
addressed. We suggest that limiting eligibility for bonus to the CHMT members alone
would be divisive. We recommend, therefore, that all health staff working for the
administration of the Council Health Department be eligible for bonus payments.



                                           - 16 -
Allocation to Individuals
The details on how a facility bonus is shared among staff will need some further thought.
It is essential that the design does not include “perverse” incentives – for example to
create “ghost staff”, or to dissuade facilities from taking on new staff (because the bonus
would be diluted). Provision would also need to be made for staff absences (whether
official or unofficial) since the remaining staff on duty would consider that absent staff
have not “earned it”.14

The system will also need to cater for the fact that facilities have very different work-
loads and staffing patterns. If the amount is fixed by facility type (eg per dispensary), the
amount of bonus will depend upon staff strength (fewer staff getting larger bonus). Such
an arrangement would be unfair to larger, busier health facilities that need more
personnel. On the other hand, it might encourage staff to work in under-staffed facilities
because the individual bonus would be higher.

If the amount is fixed per health worker, the effect of working in a larger/smaller,
under/over-staffed facility would be neutral. This would counter the possibility that the
system is seen as unfair to some facilities. It would also avoid the possibility that
facilities with smaller staff might actively deter the addition of new staff to maintain a
higher individual bonus.

On balance, we prefer this second option, on the basis of “fairness and transparency”
while also avoiding perverse incentives to reduce facility staff strength. This means that
the onus for redistribution of staff across facilities would rest with the CHMT (as it does
now). Hopefully, the CHMT bonus should provide an incentive to ensure that staff are
distributed across facilities in as productive a manner as possible in order to maximise
“whole council” performance.

Turning now to the level of payment per worker, the consensus among the team is that
this should be equal for all. The alternative – that bonus amount varies by rank or salary –
would be divisive and less transparent. It would also make the calculation of individual
bonuses more complex, more subject to error, and much more difficult to check/validate.
In interviews with staff at the council and facility levels, a majority (both senior and
junior) preferred the “egalitarian” bonus allocation to one that was linked to rank or
salary level. Only a few of the more senior staff felt that it should be based upon rank.

Performance-Payment Link
There are two main options for linking payment to performance: a) variable bonus
proportional to output; b) bonus for meeting target level of performance.

The first method would pay per unit of output (e.g. every extra child vaccinated). This
option has the merit of paying an ever greater amount the greater the output. However, it
would penalise smaller / less busy facilities, would be more complex to administer, and

14
  Note that a team bonus should also encourage peer pressure among staff to minimize absenteeism since
their collective bonus depends upon it.


                                                - 17 -
would not necessarily result in higher performance (facilities would get paid even if they
continued at current levels of output). It would also discriminate against councils with
lower output levels (e.g. due to less infrastructure, personnel, low population density,
socioeconomic deprivation or worse roads) and would undermine the resource allocation
formula. This option is therefore not recommended.

A second option is for the bonus to be paid specifically for exceeding service delivery
targets. In this case, facilities win no bonus for “business as usual” and must find ways to
increase output. This option depends upon target levels being set for specific services.

Although this option suffers the disadvantage of “all or nothing”, it does at least allow the
target to be modified from year to year to encourage higher and higher performance. 15
We recommend this method because if focuses on rewarding additional effort, because it
rewards improvement rather than absolute achievement (more difficult for some places
than others); because it is more transparent and easier to calculate, and because it will
empower facilities and managers to set target levels that are considered “ambitious but
achievable” within the local context.

Several targets would be set for every facility, according to a standard, agreed list of
indicators (see next section). The “best practice recommendation” is to have not more
than 10 indicators/targets. The main reason for this is to focus effort on key outputs and
to make the system as easy to understand and operate as possible. Thus the bonus to
facilities would depend upon how many of their targets they meet. If they meet all
targets, they get maximum bonus. If they meet only some of the targets, they get only
some of the bonus.

The simplest method would be to have (say) 10 targets, each with equal weighting. If a
facility meets 3 of its targets it would get 3/10 (30%) of its bonus. If it meets 8 targets it
gets 80% of the bonus. We do not recommend weighting individual targets. This is to
avoid endless debate among specialists as to whether one service output is “more
important” than another in public health terms. Avoiding individual target weights also
makes the scheme simpler to operate and more transparent to health workers.

Selection of Performance Indicators
A payment for performance scheme can be operated for a wide or narrow range of
services. In Democratic Republic of Congo, for example an NGO is proposing
performance bonuses only for family planning, including permanent methods (!).
Notwithstanding the fact that Norway’s primary interest is in MDGs 4 and 5, we believe
that performance measures should be selected so as to represent a “balanced scorecard”
of performance. Thus it should broadly represent the major service outputs of district
health services, encouraging and rewarding “across-the-board” improvements in service

15
   For example, if the target for an indicator is 60% coverage, a health facility receives no bonus whether
the level achieved is 25% or 59%. Conversely, no additional bonus is earned if the facility achieves 75% as
compared to 61%. The “fixed target” method is simpler to operate and bonus calculation is more
transparent. In case facilities greatly exceed their targets, these can be revised upwards the following year
to make them more challenging.


                                                   - 18 -
delivery. This approach helps to avoid the possibility that a performance bonus would
encourage the delivery of some services at the expense of others. It reinforces the need to
deliver the entire “essential basic health package”.

The selection of indicators needs to be linked as closely as possible with the actual
delivery of services. Thus, for example, it is no use increasing the frequency with which
pregnant women come for ANC if they are not receiving the essential ANC interventions.

This approach also argues for the indicators to represent actual services delivered, rather
than process or input indicators. It is no use increasing the “capacity” to deliver services
if that capacity does not actually translate into services delivered. For example, an
indicator like “every council must have a nutrition focal person” may be desirable, but it
does not guarantee that more nutrition interventions are actually delivered. Moreover, the
whole idea of performance bonuses is to focus attention on results. The implicit logic is
that staff will then identify and address the capacity constraints need to be solved in order
to deliver those results. Service-delivery oriented measures will also tend to encourage
service quality. If services are regarded as sub-standard by clients, they will not use them
– and facilities will need to address client quality concerns in order to meet their targets.
We therefore recommend indicators that measure interventions delivered (or a suitable
proxy) unless there is a compelling case to include quality or capacity aspects.

Previous debate on setting performance indicators has been handicapped by “special
interest” tendencies. Multiple stakeholders bring with them their own particular interests
and insist that they should be included in the indicators. The result is an increasingly long
list of indicators, many of which cannot even be measured. In this instance, it should be
noted that the indicators are NOT an attempt to supplant the HMIS system. The fact that
any specific indicator is not included in the list does NOT mean that it will not be
measured and reported on. To avoid the possibility that the selection of indicators
becomes an unresolved debate – or an increasingly long list – we recommend that
selection is done by a reasonably small panel of people, and that this panel includes a
majority of stakeholders from the council level. Other stakeholders should have the
opportunity to respond and comment on this panel’s recommendation. But the final
decision of the panel (after considering stakeholder feedback) should be binding.

We propose here a small number of criteria on which the selection of indicators should be
based:

1. Indicators collectively represent a “balanced scorecard” of basic health service
   delivery
2. Must make sense at facility level (eg TB or ARV treated at average dispensary or
   health centre too few to be a meaningful measure)
3. Not more than 10 indicators in total and no “sub-indicators”




                                           - 19 -
4. Indicators must be measurable using current reporting formats from facility to
   CHMTs16
5. As far as possible indicators should measure health interventions actually delivered
   rather than “patient contacts” or “capacity” to deliver a service
6. Indicators should present no risk of “adverse” incentive. E.g. rewarding C-section
   might inadvertently encourage intervention when it is not clinically necessary
7. The same list of indicators should be used for all councils and for all facilities within
   those councils (i.e. focus on basic services)17.
8. Indicators should be understandable by non-professionals so that the system enhances
   accountability to communities, village government, health facility committees,
   council administrations and council assemblies)

We have come up with a provisional list of services and specific indicators for
consideration by the “panel”. These are discussed in more detail in the next chapter.

Who Sets Targets, Verifies Performance?
There are two main considerations here. First, it will be essential that targets are set at a
level which is ambitious but achievable. If targets are set too low then a performance
bonus would not have the desired effect of raising standards. This makes it essential that
there is some kind of “check and balance” in the setting of facility-specific targets and
that these take into consideration the existing baseline performance.

Second, this check and balance should be done in a way that reinforces formal
accountabilities and strengthens the supervision system. We suggest that ideally the
“third parties” involved in target setting and performance monitoring should be:

1. Facility Level: Health Facility Committee; cascade supervisors (where applicable);
   CHMT
2. CHMT (whole council performance): DED/Planning Officer; Representative of
   Council Social Services Committee; Council Health Service Board; Regional Health
   Management Team.

This arrangement should provide the necessary check and balance in the setting of targets
whilst also reinforcing supervision and accountability.

Data quality & verification
As previously discussed, there must also be checks and balances to assure data quality
and avoid the possibility of data falsification (or even amendment of targets!).



16
   Several “preferred” indicators may not be measured in current information returns (eg Form 004). If this
is the case, we propose starting with the next best indicator that is included in the form, pending a properly
thought-through revision of the Form 004 content.
17
   For hospital services like inpatient medical care, minor and major surgery a separate set of indicators
would probably be merited. We suggest that hospital-specific measures not be included in the first wave but
be considered later if/when the scheme is to be extended to higher level hospitals.


                                                   - 20 -
We propose that the possibility of target manipulation be avoided by ensuring that
facility-specific and “whole council” targets agreed for the year are counter-signed by the
District Planning Officer (or similar). The targets should be entered onto the database that
is used for the subsequent data entry in such a way that it cannot be amended (i.e. only
designated authorised user can alter target and this leaves an audit trail on the system). In
case the indicators are expressed as rates/percentages, then the denominator (e.g. number
of children under-1) also needs to be agreed unequivocally and not be subject to
manipulation.

We also propose that a system of “internal data audit” be designed, drawing upon the
experience of the EPI program, to raise the quality of data reported and ensure that
deliberate falsification will be detected. In addition, tools for data collection could
include in-built measures to avoid unintentional errors through the use of “range limiters”
in data entry fields and identification of “anomalies” in trend data.

Calculating Performance and Payments
The system should be set up in such a way that performance against targets is
automatically tracked by a database at the council level. The payment due should then
automatically be calculated according to the targets actually met, and the number of staff
working there. This should avoid error or manipulation in the calculation of payments
due. The system would also provide an “audit trail” to verify the amount of payment due.
A payment voucher would then be drawn up in the usual way, supported by this
verification.

The present GFS18 classification offers a variety of options on where such bonus
payments should be reflected in the accounts. The main options are:

           Extra Duty allowance (code 250313)
           Honorarium (code 261106)
           Prizes & Gifts (code 260303)

We recommend that advice be sought from the Accountant General’s Office regarding
the appropriate GFS code to use, with due consideration for avoiding additional tax
liabilities.




18
     Government Financial System: the code classification for analysis of receipts and payments


                                                    - 21 -
5. Measurement in Practice
In this section we begin with a proposed list of indicators, including description of those
that can be measured using standard reporting returns currently in use.

We then go on to look at the functionality of the HMIS system at present in order to find
out whether the system could serve as a reliable basis for the collection of credible
performance information data. Based on our findings, we make a number of
recommendations on refinement and strengthening of the HMIS.

We next examine the Local Government Monitoring Database, and set out the potential
advantages of integrating the HMIS with this system.

Suggested Performance Indicators
Bearing in mind the considerations set out in the previous chapter, we propose the
following candidate services / output areas for consideration in selection of indicators.
We suggest that further service areas only be added to this list if a corresponding number
are taken out (to keep the total number to 10).

 S/N    Service                      Remarks
 1      Total OPD Attendance         Best available measure of overall workload /
                                     productivity of the facility vis-à-vis curative
                                     services. Includes all age groups, male & female
 2      Immunisation                 Key preventive service, indicator for Mkukuta
 3      Family Planning              Important contributor to maternal & child health
                                     (increase age first pregnancy, longer birth spacing;
                                     fewer unwanted pregnancies)
 4      Antenatal Care               Package of essential services to protect pregnant
                                     women & unborn children
 5      Delivery Services            Facility-based / skilled attendance essential to
                                     reduce maternal deaths. Indicator for Mkukuta
 6      Nutrition                    Major contributory factor to childhood illness and
                                     mortality rates. Indicator for Mkukuta
 7      Post-Natal Care              Neglected and under-utilised service. Essential if
                                     neonatal mortality to be reduced
 8      Under-1s                     Essential package of preventive and curative
                                     services for under-5s
 9      Under-5s                     Essential package of preventive and curative
                                     services for under-5s
 10     Malaria Control              Leading cause of morbidity & mortality

We turn now to look at specific indicators for these service areas. We have provided a list
of potential indicators, some being preferable to others (in view of the criteria previously
discussed). We also indicate which of these can be measured at present using standard
information returns from facility to council level.




                                           - 22 -
Preliminary List of Potential Indicators for Facility Performance Bonus (Quarterly)
Service / Area   Potential Indicators                  Remarks                       Measurable?
OPD               Total OPD visits (new & repeat)          Either of these a         Yes: Form 004, section 7
                  Total OPD (new only)                 reasonable measure of         (total OPD new+repeat)
                                                       total workload (curative)
Immunisation      Individual antigens                      For performance bonus     Yes: Monthly EPI returns
                  Fully immunised by 12 months         need single indicator         can measure all of these
                  DPT3                                     Best summary              <but can it measure “fully
                                                       indicator, but measurable     immunised” on quarterly
                                                       quarterly?                    basis?>
                                                           Proxy summary
                                                       indicator, measurable qrtly
Family            New Acceptors                            Preferred measure to      Yes, But only some of
Planning          Continuing Acceptors                 assess increase in uptake     these included in standard
                  Total FP acceptors                       Less preferred measure    Form 004
                  Couple-years protection                  Summary measure
                                                           More useful for
                                                       assessing pop. control than
                                                       FP demand?
Antenatal          Total ANC (1st & repeat visits)         Reasonable summary        Yes: But only available
Care               4+ ANC visits                       measure, but doesn’t          measure is total number of
                   1st visit <16 weeks                 distinguish adequacy of       ANC visits (doesn’t
                   Individual FANC components          visits or how early; nor      distinguish new/repeat;
                   All FANC components delivered       interventions actually        those who make 4+, or
                 (either as % ANC clients or % all     delivered                     how early in pregnancy.
                 expected pregnancies)                     Measures attendance
                                                       adequacy but not              Some individual
                                                       interventions delivered       components reported but
                                                           Measures early            not others:
                                                       attendance, but not             Number of PW tested for
                                                       adequacy & interventions      syphilis
                                                           Would have to choose        Number of syphilis tests
                                                       only 1 component at           that were +ve
                                                       “tracer”; leaves out other      Number received TT2+
                                                       components                      Number with at least 1
                                                           Ideal summary indicator   danger sign
                                                       if expressed as %
                                                       expected pregnancies
Delivery           Total deliveries                        Reasonable summary        Yes. Form 004, section 8 –
Services           Number of women referred            measure & should              but is conflated with
                   Maternal deaths                     encourage more facility-      measure of TBA-assisted
                   EMOC Capability (signal             based delivery                births?
                 functions)                                Risk of unnecessary         Captured in delivery
                                                       referral                      register (and book 2
                                                           Possible measure to       processing) file but not
                                                       capture effectiveness of      included in routine
                                                       EmOC (lives lost/saved)         No routine source data or
                                                           Possible measure for      reports
                                                       EMOC capability
Nutrition         Under-5s <60% WFA                        Ensures that growth       Yes
                  Growth monitoring                    monitoring done,                Already included in Form
                  Vit A supplementation                malnutrition detected and     004 section 6
                  Deworming                            incentive to improve            Ibid
                                                       nutrition. Recommended          Form 4 section 7
                                                       summary measure                 Deworming not reported



                                                     - 23 -
                                                         Important “tracer”
                                                     intervention
                                                         Important “tracer”
                                                     intervention
                                                         Measures coverage of
                                                     service-based nutritional
                                                     surveillance for under-5s
Post-Natal      Post-natal attendance/coverage          Measures patient           Yes
Care            PNC intervention components          contacts                          Jedwali 40C, of book 2
               delivered                                Preferred measure          total post natal attendance
                                                                                   ( from form F203)
                                                                                      Individual components
                                                                                   not reported
Under-1s        All under-1s registered                 Measures registration      Yes
                                                     but not services delivered.      Under-1s registered
                                                     Note growth monitoring, vit   captured in Form 4 section
                                                     A, immunisation already       8
                                                     mentioned above
Under-5s          Specific indicators (eg growth
               monitoring) already mentioned
               above. Any additional indicators?
Malaria           Pregnant woman receiving ITN           Proxy measure for ITN     No.
Control        voucher (or receiving voucher @       delivery                        Measurable from
               first ANC visit                           As above                  voucher issue records but
                  Infants receiving voucher ITN          Measure of prompt and     not currently reported
                  No (or %?) U-5s with malaria       effective treatment, under-     Ibid
               treated in first 24hrs with ACT       5s                              Not currently measured

Please note that this is a preliminary list, for illustrative purposes only. Further detailed
work would be required to examine appropriate indicators for specific services, with
reference to international best practice and in close consultation with the experts /
program heads of these various fields. There should also be further work to see precisely
what source data is available from registers and tally forms that could be used to track
specific indicators. As previously stated, no indicators should be included that are not
presently reported on through returns from facility to district – unless there is a
comprehensive revision and update of the content of the reporting formats (particularly
Form 4).

HMIS Functionality
A wide variety of data are presently collected at the facility level. The source data is
maintained in registers, before being summarised on to “tally forms” and the “facility
processing file” (Book 2). This monthly data is in turn summed and transcribed on to
quarterly reporting forms, the main one being Form 004 of Book 8. A fuller description
of the various books/registers and forms can be found at Annex 3. In addition to this
HMIS data, separate reporting systems continue to operate for a number of programmes,
most notably EPI, TB & Leprosy, HIV/AIDS, integrated disease surveillance and a few
other specialist inputs and programmes. Where these systems are reliable, they could also
be used for tracking performance against indicators.




                                                   - 24 -
Reports from health facilities go to the Council level, where they are input on to the new
HMIS software. The consolidated Council report19 is in turn transmitted to the Regional
Level who forward all Council reports to the National Level.

Without going into great detail, it is clear that information on a vast range of indicators is
potentially available. The main problem lies with the completeness and accuracy of data.
This in turn relates to a general lack of training, awareness and interest in data
interpretation and analysis; inadequate supervision and feedback on data collected; and a
number of relatively minor “system design” issues.

On the basis of our interviews and our analysis of HMIS data completeness, accuracy and
timeliness, we endorse the recommendation of the HMIS Review (HERA 2000) that
“data at the point of collection is looking reasonable to very good” but that the problems
arise during the collation and transfer of data from source books to summary forms. We
agree that the basic registers and forms do NOT need wholesale revision and we do not
recommend “tinkering” with the basic system for collecting source data. We agree with
their view that “In general efforts should focus on reducing the complexity of the tools
and systems rather than adding new information requirements.” In other words, the
challenge is making the system work rather than a wholesale redesign of the HMIS. We
agree that the key to this is in enabling demand and use of data from the lower level
upwards. We also believe that there are opportunities for refining software tools in use at
the district level in order to improve data quality, reduce workload, improve feedback, aid
interpretation and inter-face with the Local Government Monitoring Database:

     •   Provide automated quality assurance tools (eg range limiters, identify anomalous
         data)
     •   Design and implement data quality assurance, feedback & supervision systems
         and procedures
     •   Design automated data feedback forms (district – facility; region - district)
         including standard reports and graphical representation
     •   Design automated system to interpolate for missing data so that credible summary
         statistics can be generated at council, regional, national level
     •   Preserve facility identity of data on the system so that it remains possible to “drill
         down” to facility level; compare facility performance; ascertain level of data
         omission per indicator and facility
     •   Build “gateway” so that HMIS data (or a sub-set of it) can bridge directly with the
         Local Government Monitoring Database – improving the quality of data input and
         avoiding need for double entry
     •   Include pre-set denominators (total population, expected pregnancies/deliveries,
         population under-1, population under-5) on the HMIS database to avoid
         denominator errors and/or manipulation




19
  Unfortunately, this council summary report consolidates all data, so that the “facility identity” of data is
lost and it is no longer possible to ascertain the completeness of reporting on individual indicators.


                                                    - 25 -
Having said this, the most pressing and resource-intensive challenge will still be the
human one: training, mentoring, supervision etc.

In the course of this study, we conducted analysis of information returns from a non-
random20 sample of 10 health facilities in Ulanga, Kilombero, Kinondoni and Temeke.
The results are summarised in the figure below.




     •   In all cases, the basic registers for the recording of primary data were available
         and in use.
     •   4 out of 10 facilities (40%) had a close match between source data in registers and
         the “processing file” Book 2/Tally Forms. In the remaining 6 there were errors –
         to a greater or lesser extent - in the transcription of data from source register to
         book 2.


20
   The size of the sample and the “convenience” basis were constrained by the time available. Surveyed
clinics were: Ulanga District: Mahenge Hospital, Lupilo Health Center and Kichangani Dispensary;
Kilombero District: Kibaoni Health center and Michenga Dispensary; Kinondoni District: Sinza HC,
Tandale dispensary,Mwananyamala Hospital; Temeke District: Temeke Hospital, Mbagala-Rangi Tatu
dispensary


                                                  - 26 -
     •   For the 9/10 that had a book 10 available for inspection; 6 facilities out of 9 (67%)
         had correctly transferred the information from Book 2 to Book 10.
     •   Proportion of facility quarterly reports for Jan-March 2007 actually submitted as
         of July 2007 :
             o Ulanga District: 11 facilities / 35 total had submitted their reports
             o Kilombero District: 26 out of 46 had submitted
             o Kinondoni Municipality: 12 out of 24 had submitted
             o Overall reporting rate (% facilities reported within 3+ months of end of
                 quarter) for these districts 49/105 = 47%
     •   Note that CHMTs do follow up on missing reports and eventually achieve a much
         high level of compliance – albeit with delays.
     •   At the regional level (Morogoro), 80% of districts had submitted their quarterly
         report for Jan-March – although with the intrinsic data gaps described above.
     •   At the national level, we are told that all regions do report – but some may be late.
     •   Consolidated data at the district – which is transferred all the way up to national
         level - contains an unknown level of data omission. It is therefore not possible to
         translate this incomplete data into estimates by interpolating for missing data.

Another study, of 134 health facilities in Lindi and Mtwara Regions, found that 106
(73%) of facilities provided all quarterly returns to the district level and 84% of facilities
provided complete ANC data for the previous year. These rates seem to be markedly
better than those found in our “mini-study” just reported.

From our (admittedly small and non-random sample) we conclude:
   • Source data is generally recorded in the registers21
   • There are sometimes errors and omissions in summarising and transcribing this to
      monthly book & tally sheets
   • There are also errors and omissions in transferring monthly data to quarterly
      reports
   • Quarterly reports are often submitted late – and with omissions and errors
   • Compliance and timeliness of reporting from council to region, and from region
      to national levels is generally better – but still with the errors and omissions in the
      data that has come up from the facilities
   • Data are barely used at any level, reducing the incentive to provide timely and
      accurate reports and making data collection a “chore” rather than a help.
   • Supervision from higher levels focuses on reporting compliance rather than “what
      the data tells us” and data interpretation feedback is rare.

Assuming that these findings are reasonably “typical”, we expect that the HMIS data will
not, at present, be sufficiently robust to provide a basis for performance based bonuses.
However, we also conclude that the problems are largely human and systemic, including

21
  Registers are expected to be largely complete – particularly for OPD and MCH in smaller facilities.
However, anecdotal reports indicate that there are problems of omission in patient registers, with a
maximum 20% variance between lab registers and OPD registers in the same facility in Lindi and Mtwara
regions (Schellenberg D, personal communication)


                                                - 27 -
little implicit incentive (acknowledgement, feedback) to report accurate, complete and
timely data. We believe that this is largely a question of training and supportive
supervision, and that HMIS data quality could be greatly improved over a relatively short
time frame. We expect that there is probably a minority of councils whose data is already
good enough to introduce P4P. The number of such “ready” councils could not be
estimated without a more in-depth accreditation exercise. We also believe that the
attraction of joining a P4P scheme will provide a powerful incentive to improve the
performance of the HMIS at facility and council levels.

We recommend that simple accreditation criteria be put in place to qualify a council to
commence P4P. This should include, as a minimum, complete facility to council returns
for the previous year, close correspondence between Book 2 and Book 10 for the
previous 2 quarters, and close correspondence between Book 2 and source registers. Non-
qualifying facilities would be excluded from the P4P system until their data was up to
standard. We anticipate that a large minority of councils could be brought up to this
standard with an intensive HMIS support and capacity-building exercise.

Local Government Monitoring Database
This information system has been put in place under the Local Government Reform
Programme and is now recognised as the “official” source of council data. Its purpose is
to provide management information to council administrations and higher levels across
all sectors. Sector information (including health) is input once per year for individual
facilities. The facility identify of data is preserved and can be linked to
village/ward/division/council. The respective population denominators are already input
on to the system, enabling automatic generation of rates / coverage. The system has been
designed and built with the help of the University Computing Centre. The database is
open-source software. It has been designed with user-friendly input screens, standard
tabular and graphic reports, and the ability to provide comparative data across
villages/wards/councils and regions. The system is supported by an operator’s manual, a
co-ordinator’s manual, sector manuals, self-tutorial materials and training aides. It
includes a service and help-line facility to support users. All councils have already trained
LGMD focal persons as well as sector focal-persons on the use of the system. There are
apparently plans to put VSat systems in each of the councils to allow the LGMD to
operate as a wide-area network, with fast up/down-load speeds. In the meantime, councils
report to PMO-RALG by e-mail attachment. The system was rolled out in 2006,
including training of five staff from every local authority using a team of 32 facilitators.
Demonstration of the system shows that it does indeed do what it claims to do and the
user-interfaces are intuitive.

As with the HMIS, the main problem is the poor quality and completeness of data that is
put on to the system. This could largely be addressed if a “gateway” were built so that
cleaned / quality-assured data could be exported from the HMIS data system straight on
to the LGMD.




                                           - 28 -
We recommend that:
  • A gateway is built for exporting HMIS data (or a subset of this data) on to the
      LGMD. This will be essential to improve the quality of LGMD input data and to
      reduce workload – even if the LGMD is not subsequently adopted as a platform
      for operating the P4P system.
  • The set of indicators on the LGMD is revised and updated to make it more
      relevant for health management decision-making and performance monitoring at
      council, regional and national levels
  • The possibility of using the LGMD as the main platform for the P4P system is
      investigated. This has the added advantage that the system is maintained by a
      third party (Council Planning Department) and so would provide the necessary
      check and balance for data protection, verification and authorisation for bonus
      payments
  • A structured self-assessment of HMIS data integrity, checked on a sample basis
      by a third party, be carried out by the council health departments to determine
      how many are ready to be included in a pilot or phase 1 roll-out of the P4P system
  • A systematic, phased, capacity-building exercise (to be funded from Norway’s
      support for HMIS strengthening) be designed and implemented in order to:
          o Upgrade the HMIS skills of staff at facility and council levels
          o Strengthen the HMIS and improve data integrity
          o Refine and improve software platforms for data input and analysis
          o Institute a culture of data analysis and interpretation and design simple
              tools to assist in this22
          o Encourage regular feedback between levels, including comparative
              performance analysis and data interpretation
          o Review and refine the content of the Form 004 in order to generate the
              sub-set of agreed P4P indicators
          o Design and implement a gateway for exporting HMIS data to LGMD
          o Refine and improve HMIS software in order to preserve facility identity of
              data and monitor completeness of reporting by indicator.




22
  Much can be learned from the EPI program in this respect. Daily “temperature charts” and monthly
progress charts allow health workers easy ways to check if they are “on track” for their targets.


                                                - 29 -
6. Implementation Arrangements
This section describes in brief how the P4P scheme could be implemented.

Supply-Side; Demand-Side, or Both?
Referring back to earlier sections, a P4P scheme can include supply-side incentives
(awarded to health providers), demand-side incentives (awarded to consumers), or both.
On the basis of the preceding discussions, we do recommend that a supply-side incentive
scheme be implemented, whatever judgement is reached on the demand-side.

Our interviews indicate that demand-side incentives are a more controversial proposition,
particularly if the incentive is in the form of cash. We note that if it were implemented, a
demand-side incentive would only be applied for very specific services (eg to refund the
costs associated with referrals for obstetric emergencies). As such, the design and
implementation challenges for demand-side incentives are quite different from those on
the supply side. There is no intrinsic reason by supply-side and demand-side incentives
must be implemented together. A separate decision on the relevance and feasibility of
each can be reached and implementation arrangements can be done separately. Because
of positive experience elsewhere and our knowledge that some access problems are
rooted in the demand side, we recommend that a demand-side incentive be tried out on an
experimental basis.

The implementation of supply and demand-side incentives should be done in such a way
as to permit a thorough, case-control evaluation of the relative merits of each – e.g.:
Case 1: Supply Only
Case 2: Supply & Demand
Case 3: Demand Only
Control: Neither (but match other conditions pertaining)

The evidence thus generated would provide a solid, Tanzania-specific basis for decisions
to roll out the scheme to national scale and/or to modify it.

Universal, Phased or Experimental
Neither supply nor demand-side incentives can be implemented immediately on a
universal basis. The information systems are simply not good enough, and the
implementation effort would necessarily require a phased approach.

If Government is convinced of the merits of P4P on the basis of this feasibility study,
then it could opt for a Phased implementation approach, aiming to spread out to the
whole country as rapidly as possible.

If Government believes the idea has promise, but wants to “try before buying”, it would
make sense to have a pilot phase. The results of this would then be used to arrive at a “go
or no-go” decision for a national roll out.



                                           - 30 -
On balance, this team recommends the first option, including a solid “monitoring and
evaluation” component that tracks effectiveness of P4P in delivering the desired impact
on performance.

Complementary Measures to Raise Performance
Whilst we are convinced that P4P shows considerable promise as a means of raising
performance, we do not see it as a “unique” or “silver bullet” solution. It is clear that
other measures could be taken, with or without P4P, to help drive up performance. These
measures include:

     1. Include performance reporting as a mandatory requirement for routine basket fund
        reports. This would mean that every council would need to include performance
        targets with its annual plan. Quarterly performance against targets would be
        reported quarterly, along with the existing requirement for financial and technical
        reports. Even if this “requirement” was not rigidly linked to trigger release of
        funds, such a mandatory reporting obligation should improve reporting from
        council to the central level.
     2. Include performance reporting (possibly with a shorter, more output oriented list
        than the current 30+) as a standard requirement for the Joint Annual Health Sector
        Review.23
     3. Make greater use of comparative performance assessment across councils at the
        regional and national level. Even with incomplete data, this will help to build up a
        “data for decision-making” culture, provide feedback to councils on their
        performance, and inspire councils to compare themselves with their peers.
     4. Devolve a “modest” level of petty cash to the facility level by means of an
        imprest. This proposal is of sufficient importance that we describe it more fully
        below.

Imprest for Health Facilities
Presently health facilities at the district level have no financial control and no access to
petty cash. They are entirely dependent on the council health department to make
expenditures on their behalf, even if these are matters as minor as the purchase of a new
door handle, repair of a roof, or supply of basic equipment and supplies. This also applies
to revenues raised by the facilities through cost-sharing and the Community Health Fund,
which are held “on behalf of facilities” by the Council Health Department. Both these
funds and the “allocation” to dispensary and health centre cost centres in the CCHP /
budget are “virtual” funds, over which the facility and their health facility committee
effectively have zero control. The absence of any control over resources – and hence the
power to take any meaningful action – is probably the main reason that health facility
committees are largely non-functional.

Meanwhile, in the education sector, all funds for the operation of schools (in the form of
capitation grants) are devolved directly to schools, for management by the school

23
  Since this now takes place in September, it would be reasonable to assume that all data for financial year
just finished – or even the previous calendar year – could be presented.


                                                   - 31 -
committee (representing school administration and parents / community members). This
system offers a direct stake for governing committees in the running of the facilities that
serve their communities and gives them the resource control necessary to take remedial
action.

We would not go so far as to say that all operational funds should be devolved to health
facilities. But we do believe that a meaningful petty cash facility at facility level would
enable many minor problems to be rectified AND give empower the facility staff and
health facility committee to take a more active role in governing their own affairs. We
believe that the combination of staff “intrinsic motivation” (sense of duty, ethics) plus
community accountability (through health facility committee) would generally ensure
that such funds were put to good use.

Without any control over resources, it is doubtful that a P4P system could work at
the facility level. Unless facilities have at least a modicum of resource control, they will
be wholly unable to address any constraints that require resources – except by hoping that
the CHMT will respond to their pleas.

An imprest arrangement would achieve this with minimum financial risk since it needs to
be retired regularly and financial exposure cannot exceed the periodic imprest amount. It
would not require establishment of any new bank accounts or appointment of book-
keepers. It would automatically be consolidated into the health department’s accounts
when the imprest was retired. Any outstanding imprest would be very obvious to
accounts staff and audit, and would be actively followed up. To provide assurance of
financial control, the prospective imprest holders would be provided with a half-day
training on how to manage and retire the imprest.

We therefore recommend that:
  • An imprest facility for every health facility and in every council be adopted as a
      matter of national strategy
  • Imprest guidelines and basic training is designed
  • An imprest level is determined, with due consideration of the overall budget
      envelope held in a typical district and the amount that is allocated to the health
      facility cost centres
  • We do NOT recommend initiation of a P4P system unless an imprest arrangement
      is already in place




                                           - 32 -
7. Indicative Costs
The time allocated for this feasibility study could not permit a very detailed costing.
There are also a number of unknowns24. The cost of implementing P4P will clearly
depend upon the number of councils to be included at phase 1 and how rapidly the
scheme is rolled out. Below, we therefore provide only a very “broad brush” indication of
cost categories, and an order-of-magnitude estimate for each. These figures should be
seen as strictly indicative. The specific tasks and amounts would doubtless change once
the more detailed design and implementation arrangements have been decided.

The major cost categories are summarised in the table below.

Cost Element              Description               Estimate                  Remarks
Performance Bonus         Max amount per            Max performance           Funded from
Payments                  health worker to be       bonus /qtr (whole         Norad’s
                          determined.               country) approx $80       contribution to the
                          Expected to be            x 25,000 council          basket fund.
                          around per health         level health workers
                          worker 100,000/=          = $2m per quarter;        For non-
                          per quarter (about        $8m per year.             participating
                          10+% of CO salary         Expect 60%-70% of         councils, used as
                          for 3 months)             targets to be met, on     general revenue for
                                                    average: $5m - $6m        CCHP
                                                    per year if taken to
                                                    national scale
Detailed Design of        Consultations /           <$100,000                 May require
P4P scheme                consensus meetings                                  multiple meetings,
                          with MOF, PMO-                                      including regional
                          RALG, PO-PSM,                                       level
                          Regions, Councils
                          Determine desired         <$100,000                 Requires relatively
                          level of P4P bonus;                                 quick consultancy +
                          precise modalities;                                 consensus meetings
                          rules/guidelines;                                   at national level
                          indicators; design
                          self-assessment tool
                          Assessment                <$10,000                  Assessment of
                          readiness / selection                               council readiness
                          of councils for 1st                                 based on
                          phase councils                                      verification of self-
                                                                              assessments. 10-20
                                                                              councils @ $500-

24
  Whether / over what time frame to be implemented; size of maximum performance bonus; level of IT
design & development effort required; best modalities for training implementation etc.


                                               - 33 -
                                                             1,000 incl. travel,
                                                             subsistence, fees
HMIS              Training needs           <$50,000          Fees, consultations
Strengthening     assessment, training                       (venue, per diems)
(Council Level)   design
                  Develop manuals,         <$50,000          Fees, venue, per
                  guides, training                           diems
                  materials, test, adopt
                  Training master          <$50,000          Fees, venue, per
                  trainers                                   diems
                  Training of trainers     <$100,000         Fees, venue, per
                                                             diems
                  Training on HMIS,      <$2m                $50/day x2staff
                  data interpretation                        x5days x4,000
                  & use, P4P, 2 staff                        facilities = $2m.
                  per facility, <5 days,                     Paid for from
                  at council HQ                              CCHPs?
                  Software design,       <$10,000            Fees
                  refinement (HMIS,
                  LGMD gateway,
                  P4P modules)
                  Facilitator Team       $500,000           10 facilitators,
                  (follow-up,                               $20,000 each incl.
                  mentoring, trouble-                       travel & subsistence
                  shooting, system                          = $200,000 per
                  maintenance, incl.                        year; 3 years
                  travel, subsistence)
                  Ongoing IT             $200,000           $100k per year for
                  maintenance                               years 4 & 5
                  contract incl.
                  trouble-shooting;
                  refresher training
                  Hardware               $50,000            100 computers @
                  (replacement                              $500
                  computers,
                  computers for
                  hospitals)
TOTAL             $28.2m over 5 years, of which $25 bonuses; $3.2m capacity-
                  building




                                      - 34 -
8. Conclusions and Recommendations
We conclude that there is strong support among many stakeholders for the institution of a
performance-related bonus system in the Tanzania health sector – albeit with various
provisos and concerns. These concerns have already been described in full, and most if
not all of them can be addressed through careful design of the scheme.

There are already a number of precedents for performance-related pay or “in-kind”
rewards in Tanzania. However, the SASE experience has dampened enthusiasm because
it was seen as unfair, and not actually linked to individual performance.

There is a growing appetite for improving accountability for performance across the
whole of government. In a decentralised environment, where MOHSW cannot dictate
local plans and budgets, “managing for results” becomes the only way in which
performance can be driven up.

“Pay for Performance” has recently received growing attention in the health sector. It has
already been introduced (in various guises) in a wide variety of settings, including
developing countries. In fact, payment linked to outputs is the dominant form of provider
financing in insurance-based health care systems. It is also widely used in the private
sector.

Experience shows that the success of P4P schemes is critically dependent on careful
design. There are a number of very encouraging examples of P4P, including Rwanda,
where it has been rolled out on national scale. However, formal evaluations and
published results are scarce.

Our feasibility study indicates that P4P could be introduced in Tanzania. We have
carefully considered risks as well as practical considerations, and have incorporated these
into our recommendations for the design.

Two conditions are absolutely vital if P4P is to be introduced successfully: a robust
information system and a degree of budgetary authority at health facility level.

Our snapshot study confirms major problems with data quality, completeness and
timeliness. The problems mostly arise as source data from registers is compiled and
summarised for transmission upwards. Without credible data, the scheme will fail. We
therefore see a major, nationwide, HMIS strengthening initiative as an essential measure.
We argue that the HMIS problems are mainly to do with its functionality rather than its
design. We advocate a particular focus on increasing the use of data at all levels, from the
facility level upwards.

The internal logic of P4P rests on the assumption that health facility staff have the
freedom and the means to address local constraints. This will not be the case if health
facilities have zero control over resources. We therefore argue that a P4P system should


                                           - 35 -
not be introduced at the facility level unless a certain amount of budgetary control is
devolved to them. We argue that an imprest facility provides a simple and low-risk way
of achieving this. We feel that providing an imprest to health facilities is desirable in its
own right, and can be expected to bring about performance improvement, even in the
absence of P4P.

As a novel initiative, P4P should be carefully monitored and evaluated. This will be vital
to ensure that design can be adapted as problems are identified. Given the lack of
empirical evaluations of P4P in the international literature, we also argue that a formal
evaluation component should be designed and implemented alongside P4P.

Next Steps

     1. The analysis, proposals and options described here should be put before a high
        level committee that includes representation from stakeholders from Local
        Government Administrations, Council Health Departments, PMO-RALG, MOF
        and PO-PSM. This committee needs to reach decision on the key design and
        implementation options set out in this report.
     2. Appoint a national steering group to lead the process. Commission a team
        and/or contractor to undertake the detailed design work for P4P, under the
        guidance of this steering group. This will include finalisation of indicators to be
        used, level of bonus, data quality audit arrangements etc. It will also include the
        selection of “first wave” councils, based upon the robustness of their health
        information systems.
     3. Contract a team to lead and support P4P introduction in the first wave councils.
        This should include HMIS strengthening starting with the first wave and
        “rolling out” to all remaining councils.
     4. Contract separately a group to undertake scientifically robust monitoring and
        evaluation. Interested groups should be requested to submit their evaluation
        proposals for a limited tender selection.




                                            - 36 -
     Annex 1: People Consulted
Ministry of Health
   1. Mr. W. C. Mukama             PS-MOHSW
   2. Dr. D. Mtasiwa               Chief Medical Officer
   3. Ms. R. Kikuli                Director Policy & Planning
   4. Dr. Z. Berege                Director Hospital Services
   5. Dr. C. Sanga                 RCHS Country Coordinator
   6. Dr. F. Njau                  Head, Health Sector Reform Secretariat
   7. Ms. G. Minja                 HSRS
   8. Mr. J. Kelya                 HSRS
   9. Mr. M. Mapunda               Economist, DPP
   10. Dr. S. Egwaga               Head NTLP
   11. Dr. E. Nkiligi              Data Manager NTLP
   12. Dr. Alex Mwita              National Malaria Control Programme Manager
   13. Mr. J. Rubona               Head HMIS
   14. Dr. M. Kitambi              EPI Manager
   15. Dr. D. Manyanga             EPI Data Manager
   16. Ms. J. Bomani               Project Administrator, EPI
   17. Ms. A. Nswila               District Health Coordinator
   18. Dr. Amos O. Mwakilasa       Directorate Human Resources
   19. Dr. Ngonyani                Office of CMO – Quality Assurance
   20. Dr. E. Mng’ong’o            Asst. Director Private & Voluntary Health Services
   21. Dr. Georgina Msemo          Project Officer SNL-IMCI
   22. Mrs. Lena Mfalila           National SMI Coordinator
   23. Ms. Anna Nswilla            Health Secretary Preventive Dept. MOH
   24. Dr. Angela Ramadhan         National Coordinator PMTCT

Other Central Ministries and Key Informants
   25. Ms. J. Mahon                 Regional Health and Poverty Adviser, SDC
   26. Dr. H. Mshinda               Director, IHRDC
   27. Mr. G. Yambesi               Permanent Secretary, Public Service Management.
   28. Prof. Semboja                Executive Director, REPOA
   29. Mr. S. Nyimbi                Director Local Government PS-PMORALG
   30. Ted Valentine                Consultant, Public Service Management
   31. Stein Torgersbråten          Norwegian Embassy DSM
   32. Bodil Day                    Norwegian Embassy DSM
   33. Lene Lothe Gomez Palma       Health Adviser, NORAD

Dar es Salaam Region
   34. Judica Msangi               RTLC
   35. Mr Msumi                    Temeke Municipal
   36. Mrs. Lulu                   Temeke Hospital Data Manager
   37. Dr. Sylvia Mamkwe           Municipal Health Prog.Coordinator
   38. Azamah Ngwanda              Municipal Health Officer Supplies Officer
   39. Mercy Ndekero               CHAC


                                       - 37 -
   40. Mchanila F.C.            HBC-CO
   41. Dr. V. Ludovick          HIV/AIDS STI Coordinator Kinondoni M. Council
   42. Dr. Jerome Kamwela       DMO Temeke
   43. Dr. Beatrice Byalugaba   DMO Kinondoni
   44. Mr. R.D. Mutagabwa       H/Secretary Kinondoni
   45. Ms.Mwajuma Magoma        Supplies Officer K’ndoni
   46. Eugenia C. Mashoko       Accountant
   47. Boniface Tematema        Lab. Technologist
   48. Ziada Sellah             Nursing Officer
   49. Sophia Josephat          Pharmacist
   50. Alex Baguma              Principal Health Officer
   51. Dr. Emilton Ndashau      Dental Coordinator
   52. Dinah Atinda School      Health Coordinator
   53. Mary Massay              RCH Coordinator
   54. Dr. Rwechungura          Mwananyamala Hospital
   55. Dr. Rehema               Tandale Dispensary
   56. Dr. Hadija               Sinza Health Centre

Coast Region
   57. Mrs.Gertrude Mpaka       Regional Administrative Secretary, Coast Region
   58. Dr Winani                Regional Medical Officer
   59. Ms Grace Chuwa           Regional RCH Coordinator
   60. Dr. Singano              RTLC
   61. Mr. Ali Nassoro          RNO
   62. Ms. Anna Mbala           RHMT, Pharmacy Technician
   63. Dr James Malele          Ag. District Medical Officer, Kibaha (Rural)
   64. Zaina Mlongalawa         Public Health Nurse, Kibaha (Rural)
   65. Health Facility staff    Mwendapole Dispensary, Kibaha District
   66. Mr Kyombo                Kibaha (Town) CHMT
   67. Dr Kahwili               Kibaha (Town) CHMT

Morogoro Region
  68. Mr. Jackson Minja         HMIS coordinator
  69. Mr. Mgula                 District Planning Officer
  70. Mr. Chamkaga              Economist, district planning department
  71. Mr. Chief Mwakilasa       HMIS Coordinator, Ulanga
  72. Mr. Wankabale Mkessey     Acting District Health Officer, Ulanga
  73. Ms. Karerina Kaundinda    Public Health Nurse in MCH clinic, Ulanga
  74. Dr. E. Munisi             Kilombero
  75. Mr. Ndauka                HMIS coordinator, Kilombero
  76. Mr. Ntangile              District Nursing Officer, Kilombero
  77. Mr. K. Mbonja             Kilombero




                                    - 38 -
Annex 2: References
Performance-Based Financing

- -. Contracting for Health Services in Developing Countries. Discussion Draft, March
2007. <in preparation>.

Alia R. PBF Study in Uganda 2003-2005 (Preliminary Report). Presentation made at the
PBF Workshop in Kigali, Rwanda, 2007

Barros FC, Baughan JP and Vicotra C. Why so many Caesarean sections? The need
for a future policy change in Brazil. Health Policy and Planning, 1 (1): 19-29, 1986.

Beith A Eichler R Weil D. Performance-based incentives for health: A way to improve
Tuberculosis Detection and Treatment Completion? Centre for Global Development,
Working Paper April 2007.

Dudley R & Rosenthal M. Pay for performance: A Decision Guide for Purchasers.
Agency for Healthcare Research and Quality, USA, April 2006.

Dupas, P. The impact of conditional in-kind subsidies on preventive health behaviors:
evidence from western Kenya. EHESS-PSE, Paris, 2005

Eichler R. Can “pay for performance” increase utilization by the poor and improve the
quality of health services?” Discussion paper for the first meeting of the working group
on performance-based incentives, Centre for Global Development, Feb. 2006.

Eichler R, Auxilia P, Antoine U, Desmangles B. Performance-Based Incentives for
Health: Six Years of Results from Supply-Side Programs in Haiti. CGD Working Paper
April 2007.

Furth R. Zambia Performance-Based Incentives Pilot Study. Final Report. Prepared by
Initiatives Inc. for Government of Zambia and USAID. September 2005

Glassman A Todd J & Gaarder M. Performance-based incentives for health. CCT
programs in Latin America and the Caribbean. Centre for Global Dev’t Working Paper
April 2007.

Hecht R, Batson A & Brenzel L. Making health care accountable. Why performance-
based funding of health services in developing countries is getting more attention..
Finance and Development, March 2004

Kindig D. A pay-for-population health performance system. JAMA 296 (21) 2611-2613,
2006.




                                          - 39 -
McNamara P. Quality-based payment. Six case examples. Int. J. for Quality in Health
Care 2005, vol 17 (4) 357-362

Meesen B, Kashala J-P & Musango L. Output-based payment to boost staff
productivity in public health centres: contracting in Kabutare district, Rwanda. Bulletin
of WHO, Feb 2007, 85(2) 108-115.

Mohr T, Rajobov O, Maksumova Z, Northrup R. Using incentives to improve
tuberculosis treatment results: lessons from Tajikistan. Project Hope / Core Group, 2005.

Morris S, Olinto P, Flores R, Nilson E, Figueiro A. The impact of conditional cash
transfers on child weight gain. The case of the Bolsa Alimentacao Program in the
Northeast of Brazil. From “Selected Issues on Measuring and Addressing Inequities in
Health in Latin America”

Mullen K, Frank R, Rosenthal M. Can you get what you pay for? Pay-for-performance
and the quality of healthcare providers. November 2006. <citation?>

Regalia F & Castro L. Performance-Based Incentives for Health: Demand and Supply-
Side Incentives in the Nicaraguan Red de Proteccion Social. CGD Working Paper, April
2007.

Rosenthal M, Dudley R. Pay-for-performance. Will the latest payment trend improve
care? JAMA 297:7, 740-744

Rusa L, Fritsche G. Rwanda: Performance-Based Financing in Health. Third
International Roundtable. Sourcebook 2nd Edition. Working Draft.

Soeters R, Habineza C, Peerenboom P. Performance-based financing and changing the
district health system: experience from Rwanda. Bulletin of the World Health
Organisation 2006; 84:884-889.

Stedman J, McCallion G. Performance-Based Pay for Teachers. CRS Report for
Congress. January 2001.

Town R, Wholey D, Kralewski J, Dowd B. Assessing the Influence of Incentives on
Physicians and Medical Groups. Medical Care Research and Review, Vol. 61 No. 3
(supplement to September 2004).

Wisconsin Policy Research Institute. Performance-based pay for teachers in Wisconsin:
Options and opportunities. WPRI Report Vol. 14 No. 4, June 2001.

World Bank. Health Development. The World Bank Strategy for Health, Nutrition, and
Population Results. April 2007.




                                           - 40 -
Tanzania-Specific Documents

HERA. Review of the Health Management Information System (HMIS/MTUHA), Draft
Report, March 2000 (Volumes 1 & 2).

HERA. Technical review: District health service delivery in Tanzania: where are we in
terms of quantity and quality of health care provision? HERA, April 2007

Malecela, W. et al. Situation Analysis of Emergency Obstetric Care for Safe Motherhood
in Public Health Faclities in Tanzania Mainland. NIMR and Ministry of Health, April
2006.

Manongi R, Marchant T, Bygbjerg I C. Improving motivation among primary health
care workers in Tanzania: A health worker perspective. Human Resources for Health
2006, 4:6

Rubona, J. Routine Health Informaiton Systems that Operate in Tanzania. Undated
monograph.

Rubona, J. Strengthening of the Health Information Mangement System (HMIS).
Strategic Steps to Improve the Performance of the HMIS. Undated monograph.

Smithson, P. Current arrangements for performance measurement and reporting (in the
health sector). Phase 1 report, commissioned by NORAD and MOHSW (unpublished),
July 2007

Smithson P. Fair’s Fair: Health inequalities and equity in Tanzania. IHRDC &
Women’s Dignity Project, Tanzania. 2006

Smithson P. Draft Health Budget for 2007-08. A brief commentary. June 2007.

Smithson P. Local Government Reform Programme. Health Working Paper, April 2007.

UCC. LGMD Operational Manual, University Computing Centre, Dar es Salaam

UCC. Co-ordinator’s instruction manual, University Computing Centre, Dar es Salaam

UCC. LGMD Sector supervisor manual, University Computing Centre, Dar es Salaam

UCC. LGMD Rollout for the Local Government Reform Programme. Phase 1 & 2A –
Training of Trainers and Training of Local Authority Staff. Final Report, 2006.
University Computing Centre, Dar es Salaam

URT. Local Government Reform Programme. Policy Paper on Local Government
Reform. Min. of Regional Administration & Local Govt. October 1998.




                                        - 41 -
URT. National Primary Health Care Supervision Guidelines. Ministry of Health, Issue
No.1, January 1999.

URT. Tanzania Reproductive and Child Health Survey 1999. NBS & ORC Macro 1999

URT. National Package of Essential Health Interventions in Tanzania, MOH, January
2000.

URT. Performance Target Setting in the First 37 Phase Reform Districts. Mwisongo A et
al, MOH/NIMR. November 2000.

URT. Second Health Sector Strategic Plan (HSSP) July 2003 – June 2008. MoHSW
2003

URT. Proposals for Health Sector Reform. MOHSW December 2004.

URT. National Tracer Standards and Indicators for Quality Improvement in Healthcare
(Draft 1). Ministry of Health, July 2005

URT. Tanzania Demographic and Health Survey 2004-05, NBS & ORC Macro, 2005

URT. Local Government Fiscal Review 2005. Coordinating Block Grant Implementation
Team, 2005.

URT. Poverty and Human Development Report 2005. Research & Analysis Working
Group. 2005.

URT. National Strategy for Growth and Reduction of Poverty (NSGRP/ MKUKUTA)
2005 – 2010. GoT, 2005.

URT. Tanzania Service Provision Assessment Survey 2006. Preliminary Report. NBS &
ORC Macro, 2006.

URT. Annual Health Statistics Abstract. MOHSW, April 2006.

URT. Mkukuta Monitoring Master Plan and Indicator Information. MPEE, December,
2006.

URT. National Health Policy (Draft), 2006.

URT. Practitioners’ Workshop Report (on Local Government Monitoring Database).
December 2006, PMO-RALG.

URT. National Road Map Strategic Plan to accelerate reduction of Maternal and
Newborn deaths in Tanzania 2006-2010. MoHSW February 2007.




                                        - 42 -
URT. Comprehensive Council Health Planning Guideline. MOHSW & PMO-RALG,
February 2007.

URT. Comprehensive Council Health Plan (CCHP) 2006-2007 assessment report.
Petersen M, Nyaywa S. Health Sector Programme Support, MoHSW. March 2007.

URT. Mpango wa Maendeleo ya Afya ya Msingi 2007-2017. Primary health services
development programme 2007 – 2017. MoHSW, May 2007.

URT/Norway. Norway Tanzania Partnership Initiative: Programme Document, June
2007.




                                    - 43 -
Annex 3: Functionality of HMIS
Introduction
The primary source of routine service statistics is the Health Management Information
System or MTUHA. This was developed in the early 1990s, piloted in 1993 and rolled
out to all regions by 1997, including provision of computers and software to all of the
regions. A nationwide training effort took place in 1994-7 but has not since been
repeated.

Although the intention had been to fully integrate separate information systems under the
HMIS umbrella, this has not been possible – partly because of specific information
requirements, partly because the parallel systems provider better data. The major parallel
systems in operation include EPI, TB & Leprosy, Integrated Disease Surveillance and
HIV/AIDS. There is also a multitude of other systems for specialist use including
pharmaceutical supplies, onchocerciasis, trachoma, lymphatic filariasis etc, etc.

A systematic review of the HMIS was undertaken by the MOH/HERA in 2000. This
highlighted a number of (well-known) problems, including:
   • Data collected is incomplete and sometimes inaccurate
   • Incomplete and late reporting
   • Inadequate analysis and use of data for decision making at all levels – from the
       facility to the national level


Reviews and Remedial Action
Various reviews of the HMIS have been undertaken in recent years. These include:
 Title of the review and when        Motivation for the review
 was it
 R1-Minisry of Health Review         A management procedure to identify and adjust problematic
 workshop,1997                       areas and to put into account new development and
                                     technology
 R2-Assessment of infectous          Assessment of infectious disease surveillance systems in
 disease surveilace systems in       Tanzania. From this review, the Integrated Disease
 Tanzania, 1999; Joint Initiative of Surveillance (IDS) was recommended
 MoH, WHO and USAID
 R3-HERA/HMIS external Review, • Recommendation from the health sector reform
 January 2000                            secretariat and partners
                                     • Aimed to identify problems and recommend measures to
                                         strengthen HMIS
 R4- Plan of action (PoA) post       Follow up action on HERA/HMIS external Review of January
 HERA/HMIS review,November           2000
 2000
 R5- Stakeholders consultative       Recommendation from previews reviews with regard to;
 meeting on the development of       burden of work placed on primary data collectors’ emanating


                                             - 44 -
 minimum package of health         from the volume of data collected within HMIS
 information, June 2001
 R6-Health Information for         Following recommendations of the Joint health sector review
 decision making; reconciling      in April 2003, a task group on heath information systems was
 systems and approaches:           established with one of its objectives to address the need for
 February 2004                     co ordination and harmonization of health information
 R7-Health management              A management procedure to identify and adjust problematic
 Information system technical      areas
 review, June 2004
 R8-The Tanzania Health            Preliminary analysis for Health Metrics Network Initiative on
 Information system preliminary    Health Information System (HIS)
 analysis, July 2005 by Health
 Metrics Network

Partly as a result of these reviews, a number of measures have been undertaken to
strengthen the HMIS:

   •   Additional staff for the HMIS Unit in the Information and Research Unit of the
       DPP
   •   Procurement & distribution of HMIS computers for all CHMTs
   •   Development and roll-out of new HMIS software to capture, analyse and collate
       returns from the facilities, for upward transmission to Regions and National
       levels.
   •   A Monitoring and Evaluation Task Force has been convened
   •   Publication in 2006 of a Health Statistics Abstract with extensive analysis and
       interpretation of the data (the previous one having been published in 2002).


Organisation
MTUHA structure comprises of the village (community), health facility, district, regional
and the national- central level. Health facilities are the service delivery points and at the
same time generate data for the services they render. The other levels beyond that are
responsible for provision of management support and coordination to subordinate levels.
Figure 1.0 summarises MTUHA structure and at the same time illustrating flow
of resources and plans (top down) and on the other hand flows of health data from health
facility upwards.
    • HMIS unit in the planning department at MoH is the main coordinating body for
         national implementation of MTUHA
    • The ToR includes collection of information from regions and production of
         annual health statistics and other reports of national interest.
    • RHMTs report to central HMIS and co ordinate MTUHA activities in their
         regions along with providing management support to districts
    • CHMTs report to RHMTs and co ordinate MTUHA activities in their districts
         along with providing management support to HFs
    • HFs provide servces and generate data



                                            - 45 -
Inputs and logistics
  •   MoH staff establishment requires that one medical recorder is available at health
      center (HC) and hospital. In practice this is not the case in most cases especially at
      HC level.
  •   MTUHA focal person should be available at CHMT and RHMT level and this is
      true in practice
  •   Data and information management at Health Facility (HF) level is a manual
      process on papers yet
  •   At district and regional level paper based system in the form of the "District
      Processing File -DPF” work in combination with computer system. The current
      expectation is for the paper system to end at district level though districts are
      sometime sending their reports on papers instead of floppy disks.




                                          - 46 -
MTUHA TOOLS

Data TOOLS: formats- paper based & computerized
                                  Data collection tools
      HMIS BOOKS                                     FUNCTION
         Book 1              HMIS Guideline
             Book 3                Village visits register
             Book 4                Ledger
             Book 5                OPD register
             Book 6                ANC register
             Book 7                Children register
             Book 8                Family Planning
             Book 9                Diarrhea Treatment Corner
            Book 11                Oral Health
            Book 12                Deliveries
              F201                 Children attendance
              F202                 Vaccine and Vitamin A
              F203                 Routine services (HF & Villages)
              F204                 Neonatal tetanus in communities
                                         Data capturing tools
             Book 2                Health Facility Information

            Book 10                Reports from HFs to districts ( forms F001-F009 below)

Health Facility returns to DMOs ( all forms from book 10)
   TOL            REPORT TYPE            PERIODICITY                 TIME               SOURCE
 F001          Staff list                Annually                  JAN            personal details
 F002          Permanent equipment       Annually                  JAN            All equip
 F003          Buildings & facilities25 Annually                   JAN            All rooms listed
 F004          Quarterly HF statistics Quarterly                   End MAR        HMIS book 2
 F004          Quarterly HF statistics Quarterly                   End JUN        HMIS book 2
 F004          Quarterly HF statistics Quarterly                   End SEPT       HMIS book 2
 F004          Quarterly HF statistics Quarterly                   End DEC        HMIS book 2
 F005          Annual HF statistics      Annually                  End DEC        HMIS book 2
 F006          Repair & replacement ( Annually                     End DEC
               buildings & equipt)
 F008          Damaged equipment         On demand
                                   26
 F009          Infectious diseases       On demand

25
  Facilities like water, electricity, toilets, waste pit, waste water channels etc
26
  Infectious diseases are AFP, cholera, dysentery, measles, meningitis, neonatal tetanus, plague, rabies,
rabid animal bite, louse borne typhus and typhoid


                                                   - 47 -
Information Flow

Book 3    Book      Book    Book      Book   Book      Book   Book   Book   Book   Tally
           4         5       6         7      8         8      9      11     12    sheets




                                       BOOK 2
                                   (by HF in-charge)


                                       BOOK 10
                                   (by HF in-charge)



                       Computer based HMIS software: MTUHA 210
                            Data entry module
                            Report navigation module (D004 & D005)
                      (by District HMIS focal person from HF book 10)


                                      REGIONAL LEVEL:
                           Districts export DOO4 & D005 to regions
                             (by regional focal person from DPF)


                            NATIONAL LEVEL INFORMATION?
                             (by national coordinator from RPF)



HMIS in Practice
System:
   • Well established at all levels: all inputs available
   • No specific skills for data collection, analysis, use and sharing.
   • Some levels are non functional: e.g districts to regional reports
   • Huge workload (many paperwork) as facilities become bigger
   • Low priority given to HMIS by leaders at all levels
   • No enforcements of supervisions and quality control
   • No culture for using data for development
   • Low funds allocated for the HMIS system
   • Low staff motivation for collection and use of HMIS information


                                         - 48 -
   •   Non- reinforced supervision and feedback systems
   •   Low rate of trainings and follow-ups
   •   No role model (at any level?) to stimulate the data culture

Data quality & timeliness
   • Completeness and accuracy: decreases as health facilities become bigger
   • Almost always late submission (facility to council)
   • At Central HMIS, Health Statistics Abstract Report (HSAR) supposed to come
      out annually. However, for the past decade(1996 - 2006), HSAR has been
      produced only 3 times for the years 1998, 2002 and 2006 with lag of data about 2
      years behind year of publication
   • Submission of reports from regions to Central HMIS 100%
   • 60-70% completeness on contents of these reports; all regions present reports to
      central HMIS with gaps

Workload
Seemingly huge workload (paperwork) as facilities becomes bigger. However
preliminary findings from recent study in Lindi and Mtwara has indicated median time of
30 minutes on MTUHA related work in most busy day (vaccination day) at MCH clinic
with about 40% non-productive time for the health workers (Fatuma Manzi et al as part
of IPTi evaluation to answer a research question on whether or not IPTi overload the
system). Detailed assessment is suggested on MTUHA related workload




                                          - 49 -
Annex 4: Experience and Evidence on P4P
Annotated Bibliography on Payment for Performance

Dupas, P. The impact of conditional in-kind subsidies on preventive health behaviours:
evidence from western Kenya. EHESS-PSE, Paris, 2005

NGO program to provide free ITN for antenatal attenders ie conditional in-kind
incentive. ANC uptake increased by 117%, HIV testing up 84%, ANC follow-up visits up
59% compared to baseline. Method, before vs after in 3 groups of clinics: case (3);
control in same area (3); control in different area (3). Nb part of the increase due to
migration from non-prog to programme clinics – explains “a quarter” of the increase in
ANC in case clinics. “Increase” cited above is net of the transfer effect. Conclusion:
much more cost effective than unconditional distribution of ITNs in the community.

Soeters R, Habineza C, Peerenboom P. Performance-based financing and changing the
district health system: experience from Rwanda. Bulletin of the World Health
Organisation 2006; 84:884-889.

Describes experience with performance based contracts in Cyangugu, Rwanda. C, Kigali
and Butare all started 2001. Cordaid started contracting in 2002. All 24 health centres and
4 district hospitals signed contracts. Fundholder (Cordaid), communities (through
community committees), providers (health facilities and their boards). Fundholder, CBO
and district health team monitor output & quality. Contract renewal every 3 months based
on output achieved, quality and patient satisfaction. Admin costs of the fundholder was
25% of total contract value (!). Output indicators recommend <25. C also used 120
quality indicators, measured quarterly. “showed good results in terms of use of services,
financial accessibility and motivation of health staff as well as in the incorporation of the
private sector. Nb health centres free to hire, purchase own drugs, reduce user fees, pay
incentives etc.

Mohr T Rajobov O, Maksumova Z, Northrup R. Using incentives to improve
tuberculosis treatment results: lessons from Tajikistan. Project Hope / Core Group, 2005.

Provide food supplements as incentive for patients to complete treatment. “substantially
increased tb treatment completion and cure rates. Note: combined with considerable
support to TB programme including training, protocols, diagnostic procedures, lab
facilities, drug supply, fixed dose drug formulations, monitoring, managerial support,
quarterly cohort analysis, problem identification and solution. Value of food package (2-
monthly) for patient and dependents = $172 per patient (!!). Compared completion &
cure for those receiving food supplements (defined as vulnerable) and those not (and not
vulnerable). Sample sizes recognized as small. No significance tests. Completion rates for
case group about 30% points higher.




                                           - 50 -
Eichler R. Can “pay for performance” increase utilization by the poor and improve the
quality of health services?” Discussion paper for the first meeting of the working group
on performance-based incentives, Centre for Global Development, Feb. 2006.

Premise. Even with extra resources, we consider to see poor quality, low uptake, poor
outcomes. Incentives can help to address demand side, supply side or both. Demand side.
Monetary of in-kind incentive that is given conditional on defined actions (eg kids
immunized). Supply side. Material incentive designed to change provider behaviour in
terms of output and quality (through greater effort, focus on results, encourage
compliance with protocols and innovation). Ie inputs alone are not enough: only the
“intent to provide services”. From “command and control” to “contract and incentive” to
perform.

Under-utilisation by poor for many reasons including poorer service quality, higher
opportunity cost, user fees higher relative to income, health knowledge & beliefs /
perceived benefits.

Pp20-21 describes briefly a long list of examples of p4p from various countries. Many of
these large scale (ie beyond pilot). Formal evidence available on only a small number.
Many not able to control for confounding effects (investment in the health system).
Cannot conclude that p4p the most effective or most cost-effective way of achieving
desired results. But most experiments had positive outcomes. Few examples of
“perverse” outcomes.

Concerns & constraints
      Workers migrate to areas paying bonuses at expense of others
      Workers focus on rewarded tasks and neglect others
      Sustainability of payment system
      Will morale drop (more) if scheme is discontinued
      Is P4P enough without addressing other supply side constraints
      Will constraints (eg rules, rigidities) negate impact of P4P on innovative supplier
      response?
      Will “intrinsic” motivation (social / bureaucratic duty) be replaced by extrinsic
      (money)
      Beware CCT to “coerce” people to doing things they don’t want to eg sterilization
      How to verify self-reported results? Third party audit? Community verification?
      Different mix of demand and supply side required for different
      conditions/interventions?

Pp29ff documents case studies.

Rosenthal M, Dudley R. Pay-for-performance. Will the latest payment trend improve
care? JAMA 297:7, 740-744




                                          - 51 -
Focus on health care experience in USA. Half HMOs use P4P. Experience to date shows
lessons and adverse effects, hence need for careful design. Discusses pros and cons of 5
“dimensions”
       Individual or group incentive? Depends whether under influence of individual or
       group.
       Paying the right amount. Max performance bonuses for physicians average 9%
       Select high-impact measures. Eg include clinical quality measures, cost-
       effectiveness, patient satisfaction. Nb involve physicians in selection of measures!
       Some schemes link specific system improvements eg IT.
       Link reward to quality. 70% of payment schemes use minimum protocol
       thresholds. 25% pay for improvement. Many first generation reward only the best
       (decile, quartile). Increases uncertainty and fails to incentivise improvement
       among the worst. New schemes reward all who provide top quality.
       Prioritise underserved populations. Emerging interest. Little evidence.


Meesen B, Kashala J-P & Musango L. Output-based payment to boost staff
productivity in public health centres: contracting in Kabutare district, Rwanda. Bulletin
of WHO, Feb 2007, 85(2) 108-115. recommended reading.

Project funded by SIDA, implemented by HealthNet. Good paper on theoretical and
practical rationale for P4P in Kabutare. Converted fixed top-up to payment per procedure
(new consultation, TT2+ PW, new FP acceptor, fully-immunised child, assisted delivery).
Payment set so net result for average HC would be negative. Result: increase of 53% in
individual productivity plus 18% increase in staffing, total output up 80%. Min 30%,
Max 172%. Biggest improvement among the poor performers. Conclude “staff have
much more control over the production of the health centres than was previously
thought”. Risks identified. Inflation of results (independent audit); induce unnecessary
demand; deliver services that they are not equipped to deliver (sub-quality); neglect non-
targeted activities; neglect quality in favour of quantity (use complementary strategies).

McNamara P. Quality-based payment. Six case examples. Int. J. for Quality in Health
Care 2005, vol 17 (4) 357-362.

Variety of contexts and approaches have been implemented (incl developing and
developed countries), public and private, govt purchaser/insurance/employer. Major
unanswered questions are sustainability and long-term impact. Logic: pay more for high
quality, less for low quality. Examples. Costa Rica soc security allocated 2% as bonus for
high-performing hospitals. Nicaragua MOH, performance onus for achieving defined
targets (max 17% of revenue). Incl measures like re-infection rates, patient satisfaction.
Haiti. USAID-funded NGOs. 5% withheld and can be earned back, plus 5% bonus. US
Medicare. Hospitals opt for quality-performance contracts instead of previous payment
scheme (4-9 quality indicators monitored for selected procedures). Top 10% get 2%
bonus above DRG-payment. Next 10% get 1% bonus. Penalty for lowest. US. Employer
scheme conditional on installation of IT system and min staffing standards. Note: wide




                                           - 52 -
variety of “values” on what constitutes quality. Results: Haiti and Nicaragua +ve, but
both conclude rigorous piloting and evaluation to fine-tune design options.

Kindig D. A pay-for-population health performance system. JAMA 296 (21) 2611-2613,
2006.

Based on 20 evaluations with mixed results, Institute of Medicine has already endorsed
move to P4P by Medicare. Author argues for more attention to public health rather than
only clinical care. But recognize that it’s more difficult to measure population health,
devise appropriate incentives, reward diffuse responsibilities.

Hecht R, Batson A & Brenzel L. Making health care accountable. Why performance-
based funding of health services in developing countries is getting more attention..
Finance and Development, March 2004.

3 reasons cited: Growing interest in raising performance. Extra funding (linked to results)
needed for MDGs. Providers should be more accountable. Three groups of examples:
donors funding NGOs, central govt funding local govt; performance-based aid.

NGO examples: Haiti, Guatemala, Argentina, El Salvador, Nicaragua, Afghanistan.
Guatemala example: rationale to expand coverage of essential services by contracting
non-govt providers on payment-for-service basis. Central-local govt. WB project in
Brazil. Funding according to “planned” service delivery (!). Results-based aid. Examples
WB credit converted to grant (with Gates money) if results achieved. Also GAVI (TZ),
extra money (without strings) for achieving performance. GFATM also planning to
disburse based on program outcomes. Obstacles to be addressed:
    • Difficulty of measurement
    • Lack of capacity
    • “harsh/unfair” imposition on providers

Morris S Olinto P, Flores R, Nilson E, Figueiro A. The impact of conditional cash
transfers on child weight gain. The case of the Bolsa Alimentacao Program in the
Northeast of Brazil. From “Selected Issues on Measuring and Addressing Inequities in
Health in Latin America” (citation??)

Aim: boost demand for relevant services and release income constraints on feeding.
Monthly transfers to low income families with pregnant/lactating women and.or children
under 7yrs. based on committing to charter of responsibilities. $6 to $18.7 per month.
Evaluation compares beneficiaries to matched controls. Found worse weight for age and
height for age among beneficiaries! Maybe because women thought entitlement might
cease if children thrive.

Glassman A Todd J & Gaarder M. Performance-based incentives for health. CCT
programs in Latin America and the Caribbean. Centre for Global Dev’t Working Paper
April 2007.




                                          - 53 -
Reviews evaluations of CCT programs in 7 LA countries. Clear evidence on improved
utilisation. But not enough evidence on behaviour, attitudes and impact on health
outcomes. Recommend refinement of future evaluations. Recommend CCT should focus
on most poor since food a min. requirement for other developmental gains.

Beith A Eichler R Weil D. Performance-based incentives for health: A way to improve
TB detection and treatment completion? Centre for Global Development, Working Paper
April 2007.

Supply side: to health workers or whole health facility. To improve quality of diagnosis,
expand access to treatment, improve teamwork, encourage system strengthening to
improve outcomes. Note, over 40 TB programmes providing incentives – typically tied to
treatment completion. Few target poor patients only. Most target all.
Demand side. Encourage individuals to seek diagnosis, adhere to treatment.
Results: incentives make a direct contribution to higher case finding and treatment
completion. Almost all experiments showed strong positive results. Few had “unintended
effects” (mostly neutral rather than negative). Multiple changes make it difficult to
attribute improvements to incentives alone, but results encouraging. Need “careful
design” esp. distribution of food and money.

Many examples where private sector encouraged to participate in national TB program
through provision of drugs, access to training. Got better case finding and good Rx
completion. Or NGOs contracted to treat patients. Or barefoot doctors paid per referral. Ie
provider incentives mostly designed to widen the pool of treatment centres and/or pick up
patients from private sector consultations. Few examples designed to improve treatment
compliance through existing providers.

Examples and Evidence on P4P, from Eichler “Can p4p increase utilisation by the
poor and improve quality of health services?

   1. Mexico “Progresa and Oportunidades”.

5m households, 25m population. Aims to increase school attendance, improve nutrition,
utilization of health services. Conditional cash transfer depending on school attendance
and health utilization. Several evaluations of this program. Demonstrated 35% increase in
health care utilization (public), lower private sector utilization, lower private health
expenditure. Query: Just substitution public for private or net increase? What investment
in public services to meet additional demand?

   2. Nicaragua CCT Program

6,000 households in 6 municipalities. 2-monthly cash payment for attending health
education and mandated preventive health care visits for under-5s. Providers contracted
to provide services to these households $130 per hh per year. Results: uncertain (!). Too
many other confounders. More health care visits, but not much change in essential




                                          - 54 -
interventions coverage. Query: has this program increased facility visits (to quality) or
services delivered?

   3. DRC

World Bank funded. Umbrella NGOs contracted to do health system strengthening.
NGOs sign performance contracts with providers who need to meet a mix of targets
(inputs, process, services available, services delivered). World Bank funded. No results
available. Query: what does this tell us?

   4. Cambodia

3 streams. Govt + extra funds; Govt + NGO “contracted in”; NGO |”contracted out”.
Contracted did better than govt. on most measures. Query: what does this prove? Even if
all providers have equal resources, maybe NGO are less constrained by govt to address
problems. Does this simply show that contracting out yields better performance? Was
there any difference in expenditure between the three models?

   5. Haiti

>3m covered. NGO health providers put on performance contract instead of (previous)
input contract. MSH contracts 32+ providers to cover 50% of Haiti population. Contract
based on base cost (90%) plus performance-related 10%. Results: big increase in service
coverage compared to baseline. Query: No controls. Perhaps just the effect of
more/longer investment or greater accountability?


   6. Rwanda (Butare)

19 health centers contracted according to outputs (9 public 10 non-profit). Purchase
contract (with facility based on fee for service) plus Motivation Contract (bonus to health
workers for hitting targets). Result, increase (vs baseline) on all measures. Query: No
control group. What else was provided to these health centers? What happened to
background context?

   7. Guatemala.

Contract NGOs to provide services to 3m people. 88 NGOs contracted. Must meet targets
to get contract renewed. No baseline available. No results available. Query: threat of non-
renewal a “blunt instrument” and not available if government providers contracted.

   8. TB treatment compliance

Multiple examples of demand side incentives. All initiatives with quantitative results
demonstrated improvement. Shows that patient behaviours can be influenced by reward
and/or compensation of costs. Query: At what cost?



                                           - 55 -
Annex 5: Terms of Reference
                           Terms of Reference
     Feasibility study on output / result based funding mechanisms
                     in Tanzania for MDG4 and MDG5

1. Background

The feasibility of implementing performance based funding schemes is currently being
discussed among the basket fund partners, the PMO-RALG and the MOHSW, as well as
in a workshop held with MOH and The Royal Norwegian Embassy in Dar es Salaam in
April with the purpose of exploring entry points for the Norwegian-Tanzania Partnership
Initiative (NTPI).

A “Joint Statement” has been signed between the Government of Tanzania and
Government of Norway to support the reduction of maternal, neonatal and child health in
Tanzania. The Tanzania National Roadmap to accelerate reduction of maternal and
newborn deaths provides the framework for the scaling up interventions in the area of
maternal and newborn health and, in addition to the national health sector strategic plan
and MKUKUTA. The preferred option for NTPI is to deploy the new assistance through
existing, joint funding mechanisms – and to link it closely to the amplification of
progress towards MDGs through a “performance-based financing” mechanism. The
expectation is that this will combine the benefits of reduced transaction costs, national
ownership and alignment with government plans and strategies. Importantly,
performance based financing can promote innovation and results-based action at the local
level to tackle obstacles to service delivery and drive up performance in the delivery of
key health interventions.

1.1 Output/result based financing to districts in the context of the pooled basket fund

For the FY07/08, the allocations to districts from the common donor health basket fund is
increased to equal USD 0.75 per capita. Although it was originally proposed by basket
partners that the additional 25 cents would be allocated based on a performance criteria,
timing has meant that this will be a distributed according to the existing allocation
formula (population, burden of disease and remoteness). However, given the principle
that performance incentives are desirable by government and stakeholders, the
mechanisms for performance based financing to the district health basket will be
explored. Funding from the Norwegian Tanzania Partnership Initiative (NTPI) are being
planned to support the introduction of such a scheme and add flexible funding on the top
of the already proposed scheme (USD 0.75-1.00). Performance criteria will be related to
outputs/outcomes in MNCH service delivery27. However, the model may also be applied

27
  Improved MNCH service delivery is dependent on a wide set of factors where health systems issues such
as financial management, planning and logistics management, HMIS all may contribute to improve service
delivery. Performance criteria at different levels should therefore be considered.(see section on scope)


                                                 - 56 -
to a larger part of the district health financing depending the results of the feasibility
study and and discussions with MOHSW and partners.

Before being able to make further decisions in this area there is a need to explore possible
modalities and assess the best way forward with this innovative idea. It is important that
the proposed scheme is feasible, attractive and will gain support at district, provincial and
federal level.

The operationalization of such system is subject to clarifying a number of issues such as
1) What are the relevant performance criteria (programmatic and/or managerial)? 2)
Which mechanism(s) should be employed to measure the performance – ensuring
efficiency and independence? 3) Are there political implications and risks? 4) How to
establish supportive mechanisms for underperforming (and often resource poor) districts?

It is also important to notice that the CCHPs and the reporting sections are already
subject to reviews before disbursements is allowed. Any proposal must take this into
account in order to build on existing structures and not to duplicate efforts.

                                                                                        Existing
EXISTING DISTRICT HEALTH FUNDING AND                               NTPI                 Partners
PROPOSED ADDITIONAL MNCH FUNDING FOR CCHPs




                                                                           Common Donor
                        National                                           District Health
                        programs                                           basket
            Projects
                        with funding


     NGOs

                                       Additional funds (performance/output           MOF/PMO-RALG
                                       based disbursement)                            block grant for
Community funds                                                                       health
                                       Current
                                       $ 0.75 per capita (disbursed
                                       according to standard resource
                       CCHP            allocation formula as referred to
                                       above)
Others




1.2 Clarification of concepts and different types of funding schemes
Performance based and output-based financing is increasingly used in the health sector
and can tentatively be defined as
          “the transfer of money or material goods to either demand or supply side
         conditional on taking a measurable action or achieving a predetermined target”



                                                - 57 -
Performance / output based funding can be aimed to improve provision, access, quality
and utilization of health services by stimulating behaviour change either on the provider
or user side.
Depending on where one would like to stimulate the change (i.e. of district health
management teams, facility level management or individual health workers), the basic
idea is to add an extra component to the extrinsic and intrinsic motivation factors that
may already exist for delivering quality health services.
There is evidence that pay for performance schemes may contribute to improved service
provision by stimulating focused action, innovation and increased productivity. Basically
it can help change the health sector from payment for inputs to paying for results, putting
emphasis on identifying strategies to strengthen existing systems that work or introducing
new or innovative approaches.
Output / result based funding may be implemented at different levels:
   1. From global level to recipient countries, i.e.:
               Flexible funding for strategic health sector plans and their annual budgets
               conditioned on demonstrating progress on specific indicators
               Performance based grants such as the GAVI ISS rewards or GFATM
               performance based grants.
   2. From national level to sub national levels (i.e. Local Governments / District
      Health Management teams(DHMT) i.e.:
               Additional flexible funding to district comprehensive health plans based
               on satisfactory progress towards a defined set of indicators (sub-national
               data)
               Rewards based on specific outcomes
   3. From sub-national level to individual facilities (i.e Hospitals, health centres and
      dispensaries, NGOs/CBOs)
               Paying service providers according to output ( such as immunization, no
               of deliveries, ANC, number of births attended by SBA, etc), instead of
               inputs (number of trained staff, number of beds, number of ambulances,
               etc)
   4. From health facilities to health workers i.e:
               Salary bonuses to health workers according to specific outputs (attended
               deliveries, ANC visits, immunizations contacts etc)
   5. From health facility (or other management unit) to individual patients (demand-
      side financial interventions) i.e.:
               Reward women for delivering in a health facility.
               Reward women for bringing their children for immunization
               Other voucher schemes or cash transfer programs



                                          - 58 -
2. Objective

With reference to the concepts set out in this Terms of Reference, appraise the feasibility
of implementing performance based funding in the Tanzania context with the aim to
improve access, provision and utilization of quality maternal, neonatal and child health
services without compromising the delivery and use of other priority health services.
Identify key design issues and set out options for implementation.

3. Scope and guiding principles

It is suggested that the scheme(s) should follow a set of guiding principles:
         Output / result based indicators must be clearly defined and should be sensitive to
         change and should be able to capture changes in strengthening of health systems
         Already existing indicators should be used as far as possible. However, need to
         assure relevance related to MNCH.
         Strike a good balance between the proportion of the district budget subject to
         performance-based payments and not subject to performance based payment, in
         order not to destabilize planning and budgeting processes.
         Weak performing district should be identified and be subject to a diagnostic
         review if performance is not reached. Supportive actions to be identified and
         implemented. Justification for alternative funding models for these districts needs
         to be assessed.
         Scheme should be simple.
         There is a need to carry out formative and summative evaluations of scheme to
         address issues of relevance, cost-effectiveness and sustainability and identify and
         analyse those factors contributing to improved outputs and result.
         Dimensions such as rights, equity and trust should be considered carefully.

The consultancy would provide guidance to the following issues/questions:

   1. Problem analysis

The consultancy should map current financing mechanisms for district level health
services (in particular for maternal, newborn and child health services) at the different
above-mentioned levels, and identify possible areas where performance/output based
financing could have a role in improving delivery and utilization of MNCH services.

   2. Design features of possible scheme(s)

The consultancy should develop a proposal for the main features of possible schemes that
would contribute to address the problems identified above. As already stated in the
guiding principles the scheme should be guided by simplicity, implying that eligibility
criteria (if any) and potential contractual arrangements for the scheme should be kept as
simple as possible.

           a. Main actors



                                           - 59 -
The consultancy should identify the main actors in the different schemes (MOHSW,
PMO-RALG, regional administrations, district councils, district health management
teams, individual health facilities, individual NGOs/CBOs). If several schemes at
different levels are recommended the consultancy should propose measures for how to
ensure necessary links between the different levels to reach maximum effect of
scheme(s).

           b. Performance indicators

The study will provide guidance on which output / result based criteria (programmatic
and managerial) that can be used and how easy / feasible will the monitoring
performance. The consultancy should provide advise in terms of the
          Type of indicators (mix of interventions)
          Number and mix of indicators (outcome, output, process)
          Source of data

The strengths and weaknesses in each of the existing sources of data, should be reviewed
& options proposed for their strengthening.

The consultancy should also provide suggestions in terms of how the level of
performance would be linked to the size of the disbursements and how often the output
based disbursements should happen.

           c. Mechanism to measure and verify reported results and outputs

The study will provide guidance on which mechanism(s) to be used to verify the reported
results – ensuring efficiency and independence. It is crucial that the review mechanism
would take use and build on already established structures for performance reviews. The
consultancy should map already existing review mechanisms (e.g. for Local
Development Capital Grants and Review Section of CCHPs, Joint Annual Health Sector
Reviews and MKUTA reporting systems) and compare those with other alternatives such
as independent reviews (made up from i.e. academia and local population) or peer review
mechanisms. Pros and cons for each alternative should be carefully assessed.

The consultancy should also discusse methods to be used for data verification and
provide options for sanctions in case of data manipulation.

   3. Costs of supportive mechanisms

Supportive mechanisms such as managerial training and provision of agreement
templates might be needed. The consultancy should provide an overview of supportive
mechanisms needed as well a rough estimation of costs.




                                         - 60 -
   4. Pros and cons (including the political implications and risks) of implementing
      result based funding schemes in Tanzania

The consultancy will analyze the feasibility of this proposal and compare it with other
alternatives (using annual plans from health facilities, specific program targets, letter of
interests or other mechanisms etc.).

The study will identify the overriding pros and cons, including political implications
(both gains and risks) as well as strategies to overcome them the barriers and risks.

   5. Way forward

The consultancy should provide suggestions in terms of how to build understanding and
acceptance around the scheme as well as enhancing local ownership to the objectives of
the proposed scheme(s).

4. Duration

The duration of the work should be approximately 5 weeks.

The consultancy should be done in several phases:


Phase 1 :      Map the current financing and review mechanism for CCHPs and health
               facilities and discuss their strengths and weaknesses in terms of result
               focus and for providing incentives for the delivery and utilization of
               MNCH services.

               Deadline for brief report, June 30th.
               Approx. 1 week of work

Phase 2:       Continue problem analysis for remaining levels (not covered in phase 1)
               and development of proposal for the design aspects of Performance/Output
               Based Funding Scheme

               Deadline for main report July 30th
               Approx. 3 weeks of work

Phase 3:       Further refinement after feedback from the MOHSW, PMO-RALG and
               DPG health

               Final report delivered by August 20th
               Approx 1 week of work

5. Reporting




                                            - 61 -
Reference group / steering committee / management group to be convened under
government leadership to steer the design process and related consultations. The
Norwegian Royal Embassy should be represented in the steering committee.

The team of consultants will report to above mentioned steering committee. The steering
committee will oversee the progress of the work.

6. Deliverables

       Written document appraising the feasibility of implementing performance based
       funding in the Tanzania including recommendations for key design issues and
       options for implementation. Document should be not longer than 50 pages
       including an executive summary with clear recommendations of maximum 3
       pages.
       The team will conduct a consultative workshop with local government group
       representatives, MOHSW and DP Health where they will present options and
       guide the discussion on the way forward regarding the establishment of the
       scheme.
       After the consultative workshop, the consultants will write and present a final
       draft report to the steering group for comments and adjustment if required.
       The final report will be submitted not later than two weeks after receipt of
       comments from the steering group.

7. Team members

       Nationals with intimate knowledge to the process around planning, budgeting and
       financing the CCHPs.
       International expert with experience in output based/performance based funding.

8. Timeline
              Deadline for brief report for phase 1 on June 30th.
              Deadline for draft report July 30th
              Final report delivered by August 20th




                                         - 62 -

				
DOCUMENT INFO
Shared By:
Stats:
views:33
posted:3/28/2011
language:English
pages:69
Description: P4P full name of the "Proactive network Provider Participation for P2P", is an upgraded version of P2P technology, intended to strengthen the service provider (ISP) and client communication, reduce stress and operational backbone network transmission costs, and improve P2P file transfer improved Performance. Randomly selected with the P2P Peer (peer) different, P4P protocol to coordinate the network topology data, can effectively select Peer, thereby improving the efficiency of network routing.